Chapter 3: Urban Climate

Urban climate

Processes, trends, and projections

Coordinating Lead Authors: Reginald Blake (New York City, Kingston), Alice Grimm (Curitiba), Toshiaki Ichinose (Tokyo), Radley Horton (New York City)

Lead Authors: Stuart Gafﬁ n (New York City), Shu Jiong (Shanghai), Daniel Bader (New York City), L. DeWayne Cecil (Idaho Falls)

This chapter should be cited as: Blake, R., A. Grimm, T. Ichinose, R. Horton, S. Gafﬁ n, S. Jiong, D. Bader, L. D. Cecil, 2011: Urban climate: Processes, trends, and projections. Climate Change and Cities: First Assessment Report of the Urban Climate Change Research Network, C. Rosenzweig, W. D. Solecki, S. A. Hammer, S. Mehrotra, Eds., Cambridge University Press, Cambridge, UK, 43–81. Climate change and cities

~3.33 billion today to ~6.4 billion in 2050, about a 90 percent 3.1 Introduction increase. These numbers underscore the fact that urban climate is becoming the dominant environment for most of humanity . Cities play a multidimensional role in the climate change story. Urban climate effects, in particular the urban heat island This chapter presents information on four interrelated com- effect, comprise some of the oldest observations in climatology, ponents of urban climate: (i) the urban heat island effect and air dating from the early nineteenth century work of meteorologist pollution, (ii) the current climate and historical climate trends, Luke Howard (Howard, 1820 ). This substantially predates the (iii) the role of natural climate variability, and (iv) climate earliest scientifi c thought about human fossil fuel combustion and change projections due to worldwide greenhouse gas increases. global warming by chemist Svante Arrhenius (Arrhenius, 1896 ). Figure 3.2 provides a schematic of key interactions within the As areas of high population density and economic activity, cities urban climate system. may be responsible for upwards of 40 percent of total worldwide greenhouse gas emissions (Satterthwaite, 2008 ), although Teasing out the relative infl uences of these components on various sources have claimed percentages as high as 80 percent urban climate is challenging. Natural variability can occur at (reviewed in Satterthwaite, 2008 ). Figure 3.1 shows a remotely multidecadal timescales, comparable to the timescales used for sensed map of nocturnal lighting from urban areas that is vis- historical analysis and to long-term greenhouse gas forcing. ible from space and vividly illustrates one prodigious source of Another challenge is that different climate factors may not be energy use in cities. Megacities, often located on the coasts and independent. For example, climate change may infl uence the often containing vulnerable population s, are also highly suscep- amplitude and periodicity of natural variability, such as the tible to climate change impacts, in particular sea level rise. At intensity and frequency of the El Niño–Southern Oscillation the same time, as centers of economic growth, information, and (ENSO). technological innovation, cities will play a positive role in both climate change adaptation and mitigation strategies. To survey these issues, twelve cities are selected as examples. The twelve focus cities in this chapter – Athens (Greece), Dakar Urban population recently surpassed non-urban population (Senegal), Delhi (India), Harare ( Zimbabwe ), Kingston (Jamaica), worldwide and is projected to grow from 50 percent currently London (UK), Melbourne (Australia), New York City (USA), São to 70 percent by 2050 (UNFPA, 2007 ). The urban population Paulo ( Brazil ), Shanghai ( China ), Tokyo ( Japan ), and Toronto growth rate will be even more rapid in developing countries. ( Canada ) ( Figure 3.3 ) – share a range of characteristics but also In terms of absolute numbers, urban population will grow from differ in key respects. They are all large, dynamic, and vibrant

Figure 3.1: View of Earth at night: Areas of light show densely populated, urban areas. Source: NASA 44 Urban climate: Processes, trends, and projections

tends to be more difﬁ cult in cities with limited resources; typically but not always the case in developing countries. Our focus cities include examples from both the developed and developing world, with some having already taken steps to prepare for climate change and others that have yet to start. Table 3.1 gives some summary statistics on socio-economic and geographic data, and mean climate for the focus cities.

3.1.1 Effects of cities on temperature: urban heat islands

Urban areas are among the most profoundly altered land- scapes away from natural ecosystems and processes. Figure 3.4 illustrates a recent reconstruction of the verdant “Mannahatta” island circa 1609 as compared to the current landscape 400 years later (Sanderson and Boyer, 2009 ). Pondering this visually arresting reconstruction, it is not surprising that cities have altered microclimates with, among other effects, signiﬁ cantly Figure 3.2: Conceptual framework of this chapter showing major components impacting urban climate. elevated surface and air temperatures.

The elevation in temperatures is most generally explained in terms of the basic surface energy balance processes of urban areas, and act as hubs of social and economic activity. They shortwave and longwave radiation exchange, latent, sen- also feature long-term twentieth-century climate records that allow sible, and conductive heat flows (Oke, 1987 ). With respect trend detection, projections, and impact analysis. All the cities to shortwave, or solar, radiation , surface albedo refers to the selected are likely to experience signifi cant climate change this reflectivity of a surface to visible light and is measured from century, and several illustrate the infl uence of climate variability 0 to 100 percent reflectivity. The regional albedo of cities systems such as ENSO. Some of the city examples demonstrate is significantly lower than natural surfaces due to the pre- unique vulnerabilities to extreme climate events. ponderance of dark asphalt roadways, rooftops, and urban canyon light trapping. These urban features have typical With regard to differences, the selection includes a range for albedo values below 15 percent (Table 2-4 in Rosenzweig et geography and economic development levels. The various geo- al., 2006 ). This leads to efficient shortwave radiation absorp- graphic locations allow examination of climate change impacts tion. The urban skyline, with deep urban canyons, results in in multiple climate zones. Covering multiple climate zones also a greatly reduced skyview at street level and this impedes allows for the examination of the infl uence of the major natural longwave radiative cooling processes . This urban vertical climate variability systems such as ENSO. geometry further impacts winds, generally reducing ventila- tion and sensible heat cooling. The replacement of natural A range of economic development examples is important for soil and vegetation with impervious surfaces leads to greatly vulnerability studies. Adaptation planning for climate change reduced evapotranspiration and latent heat cooling. The dense

Figure 3.3: Map of cities highlighted in this chapter. The cities were selected based on socio-economic factors and the availability of long-term climate data. 45 Climate change and cities

Table 3.1: City statistics and mean temperature and precipitation for 1971–2000.∗

City Latitude Longitude Population Mean annual temperature Annual precipitation

Athens 37.9 N 23.7 E 789,166 (2001) 17.8 °C 381 mm

Dakar 14.7 N 17.5 W 1,075,582 (2007) 24.0 °C 357 mm

Delhi 28.6 N 77.2 E 9,879,172 (2001) 25.1 °C 781 mm

Harare 17.8 S 31.0 E 1,435,784 (2002) 18.1 °C 830 mm

Kingston 17.9 N 76.8 W 579,137 (2001) 27.5 °C 691 mm

London 51.3 N 0.4 W 7,556,900 (2007) 9.7 °C 643 mm

Melbourne 37.8 S 145.0 E 3,806,092 (2007) 15.7 °C 652 mm

New York City 40.8 N 74.0 W 8,274,527 (2007) 12.8 °C 1181 mm

São Paulo 23.5 S 46.4 W 11,016,703 (2005) 19.5 °C 1566 mm

Shanghai 31.5 N 121.4 E 14,348,535 (2000) 16.4 °C 1155 mm

Tokyo 35.7 N 139.8 E 8,489,653 (2005) 16.2 °C 1464 mm

Toronto 43.7 N 79.0 W 2,503,281 (2006) 7.5 °C 793 mm

∗Population data for all cities are from the United Nations Statistics Division Demographic Yearbook, United Kingdom National Statistics Ofﬁ ce, and Census of Canada. Annual temperature and precipitation statistics are computed for all cities using data from the National Climatic Data Center Global Historical Climatology Network (NCDC GHCNv2), UK Met Ofﬁ ce and Hadley Centre, Australian Bureau of Meteorology, and Environment Canada.

Figure 3.4: Manhattan-Mannahatta: on right is a reconstruction of Manhattan Island circa 1609 (called “Mannahatta” by the Lenape native Americans), as compared to today, based on historical landscape ecology and map data. Sources: Markley Boyer / The Mannahatta Project / Wildlife Conservation Society and the aerial view of modern Manhattan, Amiaga Photographers 46 Urban climate: Processes, trends, and projections impervious surfaces with high heat capacity create signifi- Balling and Brazel ( 1987 ), Bornstein and LeRoy ( 1990 ), Jau- cant changes in heat storage and release times as compared to regui and Romales (1996 ), Selover (1997 ), Changnon and natural soil and vegetated surfaces. Westcott (2002 ), and Shepherd et al. (2002 ) have shown evidence that corroborates the earlier fi ndings of enhanced pre- There are additional atmospheric and heat source processes cipitation due to urbanization. However, the hypothesis has in cities that interact with these energy balances. Aerosols tend been disputed, and even challenged, by other data studies that to reduce the amount of incoming solar radiation reaching show no local effect on precipitation (Tayanc and Toros, 1997) the surface (a net cooling effect), while elevated ambient or even defi cits in precipitation that accompany urbanization urban carbon dioxide levels may further reduce net radiative (Kaufmann et al., 2007 ). cooling. Recent studies by Burian and Shepherd ( 2005 ) and Shepherd The high density of population and economic activity in ( 2006 ) point specifi cally to increases in downwind rainfall due urban areas leads to intense anthropogenic heat releases within to urbanization. Although topological effects may be partially small spatial scales. These include building heating and cooling responsible for this fi nding in the Shepherd ( 2006 ) case, Chen systems, mass transportation systems and vehicular traffi c, and et al. ( 2007 ) supports the Shepherd ( 2006 ) study. Shepherd commercial and residential energy use . Anthropogenic heat et al. (2002 ) and Simpson (2006 ) have defi ned an “urban rain- emission has been well documented and researched in developed fall effect ,” which is defi ned as the impact of urban centers on countries as a major factor causing the heat island phenomenon enhancing downwind and peripheral rainfall. However, there (Ohashi et al., 2007 ). As economic development, urbanization, is again no consensus for a unifying theory for the urban rain- and population growth continue in the developing countries, fall effect. Various explanations of why urbanization positively anthropogenic heat has increased there as well (Ichinose and impacts convection have evolved over the years (Shepherd, Bai, 2000 ). Growth in urbanization increases energy demand in 2005 ). The arguments include sensible heat fl ux enhancement, general and electricity demand in particular. urban heat island-induced convection, the availability of more cloud condensation nuclei in urban areas, urban canopy alter- In analogy with the well-established urban heat island, it is ation or disruption of precipitation systems, and increased sur- tempting to defi ne additional atmospheric urban “islands” such face roughness convergence. as rainfall islands , and relative humidity island s, which refer to potential urban alterations of precipitation and reductions in 3.1.3 Climate change impacts on the urban heat urban soil moisture availability due to impervious surfaces. In island processes some cities, there may be good evidence for their existence and effects, such as in Shanghai (Section 3.1.3.1). In general though, Climate change may well modify the urban heat island and the case for additional urban atmospheric islands , such as rain- rainfall effects, but the quantitative extents are unknown at this fall islands, is not as straightforward as the heat island and needs time. This is a ripe area for potential future climate research. further research and characterization. For example, warmer winter temperatures may decrease energy combustion for heating, but also increase summertime air condi- There can be little question, however, about the broad array of tioning needs, thus affecting these anthropogenic heat sources. quality-of-life issues that are generally negatively impacted by Higher temperatures in summer are likely to lead to higher levels excess urban heat. These include extreme peak energy demands , of pollutants such as ozone . Changes in temperature, precipita- heat wave stress and mortality risk, air quality deterioration, sea- tion and ambient levels of carbon dioxide will all impact local sonal ecological impacts including thermal shocks to waterways vegetation and ecosystems with effects on urban parks and veg- following rain events and impacts on urban precipitation. etation restoration, which is an important adaptation strategy . Climate-induced changes in winds could also impact urban cli- 3.1.2 Effects of cities on local precipitation mate. It should also be noted that urbanization often increases the impact of a given climate hazard. For example, a precipi- There is a longstanding interest in the question of urban tation event is more likely to produce fl ooding when natural impacts on precipitation both locally and regionally. Although vegetation is replaced with concrete , and temperatures above there is evidence that urbanization and precipitation are posi- a certain level will cause greater mortality when air pollution tively correlated, a consensus on the relationship has not yet levels are high. been reached. Early studies by Horton ( 1921 ) and Kratzer ( 1937 , 1956 ) provided indications that urban centers do play 3.1.3.1 Shanghai: many urban environmental islands a role in strengthening rainfall activity. Studies by Landsberg ( 1956 ), Stout ( 1962 ), and Changnon ( 1968 ) discussed the extent Due to its land use patterns (Figure 3.5) and high densities to which urbanization may induce and strengthen precipitation. of population and buildings, Shanghai experiences perhaps The strongest argument used in those studies to substantiate fi ve interrelated microclimatic impacts, which we frame as the role of urban centers on rainfall was the enhancement of metaphorical “islands” in analogy with the heat island: the heat downwind rainfall. Landsberg ( 1970 ) and Huff and Changnon island, dry island , moisture island , air pollution island , and rain (1972a ,b) used observed data to support this hypothesis. island. 47 Climate change and cities

The Shanghai heat island appears from afternoon to midnight The second and third Shanghai islands can be referred to under certain weather patterns. According to ground observa- as the dry and moisture island effects. The elevated inner city tion records from the past 40 years, there is an apparent mean temperatures and greater soil moisture availability in rural areas annual temperature difference of 0.7 °C between downtown and should result in lower urban relative humidity. Annual average the suburban areas (16.1 °C and 15.4 °C, respectively). The cor- data from Shanghai conﬁ rm this effect ( Figure 3.5 ). However, responding mean annual extreme maximum temperatures are the daily cycle of urban–rural humidity can be more complex 38.8 °C and 37.3 °C, respectively. This urban warmth tends to due to differences in dewfall, atmospheric stability, and freezing, appear from afternoon to midnight in mid autumn and in early which can dry out rural air overnight and lead to greater urban winter, summer, and spring under clear conditions with low humidity at night (Oke, 1987 ). This can lead to a cycle of relative winds ( Figure 3.5 ). dry and moisture islands that alternate during the day and night.

Temperature (°C) Mean Annual Relative Humidity Industrial and Warehouse Land Green Space 30.41-32.22 Roads, streets & Square Public Facilities 32.23-32.78 82 32.79-33.26 81 Transportation Arable Land 80 33.27-33.67 79 Public Infrastructure Forestry Land River 33.68-34.05 34.06-34.43 78 Residential Land Garden Water Body 34.44-34.91 (Data Source: Monitoring Data of Constructive village & Livestock & Poultry Land Under and 34.92-35.57 Small Town Land Breeding Area Before Construction 35.58-36.90 1960 to 2001)

Mean Annual Precipitation (mm) 1160 1140 1120 1100 AOD 1080 (Data Source: Monitoring Data of 0.45 0.55 0.65 0.8 0.95 1.05 1960 to 2001)

Figure 3.5: Shanghai urban islands effects. From top left to bottom right are Shanghai’s land use pattern, temperature, relative humidity, optical depth, and precipitation. For all the climate variables, a distinct pattern emerges in the area of greatest development. Source: Jiong (2004). 48 Urban climate: Processes, trends, and projections

These differences in relative humidity may have implications for extremes associated with the modes. However, many cities, espe- human comfort indices during heat waves and cold periods. cially in the developing world, lack capacity to make forecasts on their own. It must also be kept in mind that these modes of The fourth island is an air pollution island, since urban air variability are only responsible for a portion of the temperature quality is poorer than that of suburban and rural areas. The inver- and precipitation variability experienced by cities. It is there- sion layer over the urban heat island holds back the diffusion of fore not uncommon for cities to experience climate anomalies atmospheric pollutants, increasing pollution levels locally. This of opposite sign to what the particular phase of the mode would in turn leads to acid rain . In 2003, the average pH value of pre- suggest. For example, in Zimbabwe during the El Niño year cipitation in Shanghai was 5.21, with a percentage of acid rain of of 1997–1998, drought conditions were predicted, and policy 16.7 percent. In the downtown area, where industry, commerce, decisions were made based on the expectation of drought and traffi c, and residents interact closely, air pollution is severe, as reduced maize yields. However, rainfall ended up being above shown in Figure 3.5 . normal, which reduced perceived credibility of climate forecasts (Dilley, 2000 ). Conveying subtleties such as probability, uncer- The fi fth Shanghai urban island is termed the urban rainfall tainty, and risk management to stakeholders around the globe effect . According to ground observation records in the fl ooding can be a major challenge. season (May–September) and non-fl ooding season (October– April in the following year) over the period 1960–2002, the One lesson is that intra-regional rainfall patterns in these El central city experiences greater precipitation than the outlying Niño (and other modes) impacted areas can be very complex regions (urban rainfall effect), with an average precipitation that (Ropelewski, 1999 ). Another is that, while climate models of is 5–9 percent higher than in the surrounding regions. There are projected impacts are improving, achieving accurate seasonal a number of hypothetical scenarios that may produce an urban climate predictions months in advance is still a research goal. rainfall effect: (1) the urban heat island effect may contribute to In addition, with global climate change , patterns of natural cli- the rising of local air currents, which help to develop convec- mate variability themselves may change (in terms of strength, tive precipitation; (2) reduced urban windspeeds may slow the frequency, and duration of the modes). Even if the modes remain movement of storm systems over urban areas and could lengthen the same, climate change may alter the teleconnection patterns the duration of rain events; (3) aerosol pollutants may provide that drive regional climate impacts. potential rainfall “nucleation sites” in low-forming clouds over the city. Thus, the urban heat island, the urban wind regime, and We highlight two specifi c modes that have large impacts on the urban pollutant island may all be partially responsible for the temperature and precipitation in many cities around the world: the higher precipitation amount in the central city and the leeward El Niño–Southern Oscillation and the North Atlantic Oscillation. area ( Figure 3.5 ). Additional modes are briefl y described, followed by a discussion of interactions between climate variability and climate change.

3.2 Natural variability of the climate system 3.2.1 El Niño–Southern Oscillation

Among climate oscillations, the coupled atmosphere–ocean Cities are vulnerable to modes of natural climate variability phenomenon, known as the El Niño–Southern Oscillation (that is, not caused by climate change), or preferred oscillations (ENSO), impacts the most cities worldwide. The oceanic mani- in atmospheric circulation. These preferred oscillations tend to festations of this mode, known as El Niño episodes, are mainly recur at seasonal to multi-year timescales, have somewhat pre- characterized by the warming of the tropical central and eastern dictable impacts on temperature and precipitation in specifi c Pacifi c. La Niña tends to exhibit the opposite cooling effects. regions, and impact large portions of the planet. The main atmospheric manifestation, known as the Southern Oscillation, is a seesaw of the global-scale tropical and sub- Modes of climate variability centered in one location are tropical surface pressure pattern, which also involves changes able to impact temperature and precipitation in distant regions in the winds, tropical circulation patterns, and precipitation through a process called teleconnections , whereby wave (Trenberth and Caron, 2000 ). motions transport energy outward from climate variability source regions along preferred paths in the atmosphere (Hurrel ENSO is measured by several indices, including the surface et al ., 2003 ). Natural climate variability can cause signifi cant pressure difference between Darwin and Tahiti (Southern Oscil- impacts, for example by infl uencing the frequency and inten- lation Index, SOI) and the sea surface temperatures in some sity of extremes of temperature and precipitation. It is, there- selected regions in the central and eastern equatorial Pacifi c fore, important for stakeholders to recognize these risks, which ( Figure 3.6 ). El Niño episodes occur about every three to seven may be overlooked when assessing climate hazards associated years, but their frequency and intensity vary on inter-decadal with long-term global change. timescales.

Modes of natural climate variability have some level of fore- There are tropical–extra-tropical teleconnections ( Figure 3.7 ) cast predictability, which can help cities prepare for climate caused by energy transfer from the tropics , which also cause 49 Climate change and cities

Figure 3.6: Southern Oscillation Index. The SOI is one measure of the El Niño–Southern Oscillation, a pattern of natural climate variability. Red indicates El Niño episodes, and blue La Niña episodes. Source: Trenberth (1989), with data from UCAR/NCAR. circulation, precipitation, and temperature variations in sub- southeastern South America , impacts are strongest during the tropical and higher latitudes, such as the Paciﬁ c–North American austral spring (Grimm et al. , 2000 ). Opposite effects may occur (PNA) and Paciﬁ c–South American (PSA) patterns. These tel- in different periods of an ENSO episode, and thus annual anoma- econnections are associated with precipitation anomalies (depar- lies can be weak even if strong anomalies happen during par- tures from long-term averages) in cities in South and Southeast ticular months or seasons (Grimm, 2003 , 2004 ). For example, Asia , Australia , southern Africa , the southern United States , and in Jakarta , Indonesia , reduced precipitation in the fall of El tropical and subtropical South America . Niño years is often partially offset by enhanced precipitation the following spring (Horton, 2007 ). The impacts of ENSO vary by season in different regions of the world, and therefore in different cities. For example, in Box 3.1 describes the impacts of ENSO in Rio de Janeiro.

Figure 3.7: ENSO Teleconnections. Teleconnections for both warm (El Niño) and cold (La Niña) episodes are shown. The variation of the teleconnections by season for each episode is also included. Source: NOAA NCEP Climate Prediction Center 50 Urban climate: Processes, trends, and projections

[ADAPTATION] Box 3.1 Climate-pr ooﬁ ng Rio de Janeiro, Brazili

Alex de Sherbinin combined with a lack of drainage, has continued to result in ﬂ ooding during the summer rainy season (January–March). CIESIN, Earth Institute at Columbia University The few remaining lagoons, mangroves, and marshes have been affected by land ﬁ ll and sedimentation, reducing their Daniel J. Hogan absorptive capacity during extreme rainfall events. Low-lying State University of Campinas, Brazil areas around Lagoa de Tijuca and Lagoa de Jacarepaguá will largely be submerged with sea level rise of 0.5–1m.iii

As fl ooding and landslides in April 2010 and January 2011 The city receives higher than normal precipitation during the have demonstrated, Rio de Janeiro and the surrounding summer months of some El Niño events, but the connection region continue to be at high risk of climate impacts. Approxi- with ENSO is not consistent. In the summer of 1998, in the mately 200 deaths were attributed to the April 2010 fl oods end of an El Niño episode, the city was affected by severe and landslides, and several thousand people were made fl oods as a result of two intense periods of rainfall in early Feb- homeless, while 450 died in mudslides in the state of Rio de ruary that produced a total of 480 mm of rain, which consti- ii Janeiro in early 2011. Efforts will need to be made to “cli- tutes one-third the annual average rainfall of 1,200–1,500 mm mate proof” the metropolitan region, increasing its resilience (depending on location). The fl ooding in early April 2010 was to fl oods induced by climate change and variability. precipitated by 288 mm of rainfall in a 24-hour period. These fl oods were associated with a weak El Niño event, but the The city and metropolitan region of Rio de Janeiro have popu- fl oods in early 2011 were actually associated with a La Niña lations of 5 million and 11 million, respectively. Guanabara event, so the linkage is not consistent.iv Bay, where Rio is situated, faces almost due south. Rio’s dramatic topography has made it prone to landslides and Rio’s peculiar physical setting, and the circumscribed nature fl ooding in low-lying areas (Box Figure 1). With the Atlantic of suitable building sites, has spawned two kinds of response. rainforest having been stripped away from many hillsides, the One is the construction of high-rise apartments close to the thin soils have become prone to landslides, and the granite coastline (e.g., Copacabana, Ipanema and Leblon) and in and gneiss bedrock has been left exposed to weathering, fl ood-prone areas further inland; the other is unregulated con- making it more prone to decomposition and erosion. struction on steep slopes, particularly on the Tijuca mountain range. The unregulated construction of favelas (shanty towns) The coastline in this area was characterized by lagoons, has a long history, and stems from the invasion of both private estuaries, and low-lying coastal marshes, many of which and public urban lands by poor urban squatters who become have been fi lled in. The fl at topography of low-lying areas, de facto (and in some cases de jure) owners of plots of land.v

Box Figure 3.1: Rio de Janeiro: ﬂ ood and landside risk (Global Risk Data Platform Preview). Flood frequency is the expected average number of events per 100 years. Landslide frequency is the expected annual probability and percentage of pixel of occurrence of a potentially destructive landslide event ×1,000,000. Sources: UNEP/DEWA/GRID-Europe. 2009. Flood frequency. In: ISDR (2009) Global Assessment Report on Disaster Risk Reduction. United Nations, Geneva, Switzerland; International Centre for Geohazards (NGI). 2009. Frequency of landslides triggered by precipitations. In: ISDR (2009) Global Assessment Report on Disaster Risk Reduction. United Nations, Geneva, Switzerland; European Space Agency. 2009. Globcover for South America (global legend).

51 Climate change and cities

In terms of human vulnerability, Rio’s income distribution is to the narrow coastal lowlands, where they join canals whose highly skewed. The existence of pockets of extreme poverty limited fl ow capacity causes frequent fl ooding. By contrast, side-by-side with affl uent neighborhoods is characteristic of the Baixada Fluminense, a large marshy lowland somewhat the city. The 2000 census reported that 1.1 million people live removed from the steeper parts of the city, has had reason- in favelas, or 20 percent of the municipality’s population. The ably adequate drainage since the 1930s.vii city is also densely settled: the municipality has an average Generally, precipitation extremes are expected to increase density of 4,640 persons per sq km but densities in the smaller in severity with climatic change, and these will have administrative units of the metro area are between 8,000 and adverse impacts on Rio, given that the city already experi- 12,000 persons per sq km. ences extreme fl ooding and landslides on a roughly 20 year Rio has a large migrant population from the poorest parts of basis. Poor neighborhoods are particularly vulnerable to this Brazil’s arid northeast region. The fact that many of them do extreme precipitation: roughly 300 people died and more than not have personal experience with mudslides or mass wasting 20,000 people were made homeless during a 1967 fl ood, and may account for their building practices. Migrants move up in 2010 the Morro do Bumba slum built on a former garbage hillsides in search of new land, consistently eating away at dump in Niteroi collapsed and slid downhill, burying homes the vegetation cover on the slopes above the favelas, despite and killing more than 200 people. Extreme and unpredictable government efforts to cordon off such areas to prevent fur- rainfalls and fl oods converge with projected sea-level rise to ther development. New regulations have been put in place increase stresses that will be diffi cult for Rio to handle owing that restrict building in hazard-prone areas. Efforts to “reg- to the city’s topography (narrow coastal shelf backed by ularize” favelas have also been underway for several years, steep mountains subject to mass erosion), poor building con- with various government programs to undertake cadastral ditions, the lack of secure land tenure for a notable portion surveys, grant deeds to de facto owners and provide basic of the city’s population, poverty coupled with large income infrastructure. These same plans limit the further expansion of inequalities, and large problems with sanitation systems and favelas in fl ood-prone or steeply sloped areas. Rio de Janeiro sewage disposal. has invested more than US$600 million in its Programa Favela Little in the way of concrete fl ood protection infrastructure has Bairro to improve access to basic infrastructure, health and been set up in the wake of the 1988 fl oods. It is possible to education for half a million of its poorest residents.vi In terms speak of highly vulnerable sub-populations living in favelas and of social cohesion, the favelas do have some rudimentary near waterways, and relatively less vulnerable upper classes organization, including neighborhood watches and self- living in high-rise apartments in locations less susceptible to improvement societies. These can be important for self-help inundation. This speaks to the need to upgrade slums, limit and early warning systems. The drug traffi c entrenched in settlement on steep slopes and unstable locations, relocate many favelas (and the violence it spawns) is a major counter- some settlers on some slopes, revegetate hillsides, create force to these elements – a problem that has no end in sight. more green verges near waterways for water absorption, and Although favelas have always suffered during rainy seasons, improve drainage systems in low-lying areas. Part of this will the development up slope and paving of walkways has had entail cleaning and maintenance of existing waterways and the effect of increasing runoff to the low-lying areas. Runoff is canals. A number of more specifi c suggestions for climate channeled down cemented and quasi-natural watercourses proofi ng are provided in the volume by Gusmão et al. (2008).

iThis is dedicated to Daniel J. Hogan, who passed away in 2010. Dan was a leader in the fi eld of population-environment studies and did much work on urbanization impacts on the environment in Brazil. ii “April 2010 Rio de Janeiro fl oods and mudslides”, Wikipedia, accessed on January 19, 2011 at http://en.wikipedia.org/wiki/April_2010_Rio_de_Janeiro_fl oods_and_ mudslides; “Brazil mudslide death toll passes 450”, CBC News, accessed on January 20, 2011 at http://www.cbc.ca/world/story/2011/01/13/brazil-fl ood-deaths.html. iiiGusmão, P. P., P. Serrano do Carmo, and S. B. Vianna (2008). Rio Proximos 100 Anos. Rio De Janiero: Instituto Municipal de Urbanismo Pereira Passos. ivNational Weather Service’s Climate Prediction Center for a list of El Niño and La Niña years, accessed on January 20, 2011 at http://www.cpc.ncep.noaa.gov/ products/analysis_monitoring/ensostuff/ensoyears.shtml. vFernandes, E. (2000). The legalisation of favelas in Brazil: problems and prospects, Third World Planning Review, 22 (2), 167−188. viUN HABITAT (2006). State of the World’s Cities, London, UK: Earthscan. viiCunha, L. R. and M. Miller Santos (1993) The Rio reconstruction project: the fi rst two years. In Towards A Sustainable Urban Environment: The Rio de Janeiro Study, World Bank Discussion Paper 195, Washington, DC, USA: World Bank.

3.2.1.1 São Paulo and the El Niño–Southern Oscillation peak rainy season (January–February). Enhanced precipitation in the early rainy season, which is more frequent during El Niño The ENSO can have a strong effect on rainfall in São Paulo, episodes, can have a signifi cant impact, since in this case urban Brazil , although this city is situated between regions of oppo- fl oods and landslides may become more frequent during the rainy site effects of ENSO in the rainy season, Central-East and South season. Flooding and landslides are disruptive, costly, and deadly . Brazil. In Central-East, ENSO affects the frequency of extreme daily precipitation events in opposite wasy in the early and peak 3.2.2 Effects of the North Atlantic Oscillation on cities summer rainy season (Grimm and Tedeschi, 2009). In São Paulo, the behavior is different: there is signifi cant impact in the early The North Atlantic Oscillation (NAO) is the dominant pattern rainy season (October–November) but no consistent impact in the of atmospheric circulation variability in the North Atlantic region 52 Urban climate: Processes, trends, and projections

Figure 3.8: NAO Index. One measure of the NAO Index is the difference in sea level pressure between the polar low near Iceland and the subtropical high near the Azores. Red indicates the positive phase of the NAO and blue indicates the negative phase of the NAO. Source: Hurrell (1995), with data from NCAR.

(Hurrell et al. , 2003 ). Although concentrated there, the impacts NAO and climate in London is that the positive phase brings on climate extend over cities in a much larger region, from cen- wetter and warmer conditions, while the negative phase brings tral North America to Europe , Asia , and Africa. When the NAO drier and colder conditions (Wilby et al., 1997 ). In early 2009, index is positive, there is a higher than usual subtropical high London experienced a rare heavy snow fall, which crippled the pressure (located in the general area spanning Portugal and the transportation system and forced many businesses to close. At Azores) and a deeper than normal Iceland low pressure (Visbeck that time, the NAO was in the negative phase, which helped et al., 2001 ; Figure 3.8 ) This larger pressure difference pro- transport cold air to the city. The snowfall in 2009 is an example duces stronger-than-average westerly winds at middle latitudes, of how expected relationships of natural climate variability do resulting in warm and wet winters in eastern USA and most not always hold. While the colder temperatures were typical for of Europe and cold and dry winters in Canada and Greenland . the negative NAO, the increased precipitation was not . Drier and colder conditions also prevail over southern Europe, the Mediterranean, and northern Africa . The oscillation and its 3.2.3 Other modes of natural climate variability aforementioned impacts are strongest in winter, but are present throughout the year. The NAO index varies from month to month In addition to ENSO and the NAO, there are other modes of and from year to year; however, the long-term average tends to natural climate variability that impact cities. These include the remain in one phase for multi-year periods. Pacifi c Decadal Oscillation (PDO) , the Southern Annular Mode (SAM), the Atlantic Multidecadal Oscillation (AMO), the Indian The negative NAO index refers to a weaker and eastward- Ocean Dipole (IOD), and the Madden–Julian Oscillation (MJO). displaced subtropical high, and a weaker Iceland low, leading to The PDO is a longer timescale (decadal compared to inter- anomalies that are largely opposite of those associated with the annual) pattern of variability that affects the climate in the Pacifi c positive NAO index. The mid-latitude winds are weaker and in Basin. The SAM is a mode of extra-tropical variability in the a more west–east direction during negative NAO events. Warm Southern Hemisphere that affects the climate in South America and moist air is brought into southern Europe and cold tempera- and Australia . The AMO is a mode of natural variability of North tures prevail in northern Europe and in eastern USA. Canada and Atlantic Ocean sea-surface temperatures that infl uences Atlantic Greenland experience higher winter temperatures. Basin hurricane activity. The Indian Ocean Dipole (IOD) is a coupled ocean–atmospheric mode centered on the tropical Indian The impacts of NAO on precipitation and temperature in North Ocean Basin. The Madden–Julian Oscillation is an intra-seasonal America and Europe have urban consequences for hydropower mode of variability that primarily infl uences precipitation in the generation (through water availability), energy consumption tropics . (through temperatures and humidity), water supply, and ecosystems (drought and forest fi res, for example) among other sectors. 3.2.4 Natural variability and global climate change Cities highlighted in this chapter that are impacted by the NAO include New York City, Toronto, London , and Athens . A London The frequency and intensity of the modes of natural climate example is presented below. variability and their associated teleconnections may change with global climate change. Because of the non-linearity of the cli- 3.2.2.1 London and the North Atlantic Oscillation mate system, it is not possible to simply add the teleconnections of a particular mode to a changing background climate (Hoerling One city featured in this chapter that is affected by the NAO et al., 1997 ). It is important to analyze long-term observations of is London. Changes in the phase of the NAO have an impact on the modes to learn how they have varied in the past and if they are temperature and precipitation in the city, especially during the exhibiting any trends in the current climate. Climate change and wintertime. The general relationship between the phase of the climate variability may interact in complex and unpredictable 53 Climate change and cities ways. For example, shifts in mean wind patterns (such as the jet 3.3.1 The climate system and global climate stream) associated with climate change may modify the regional change teleconnections of the modes. It is possible that the accuracy of seasonal forecasts may change in some regions as climate change The elements of the global climate system include the atmos- accelerates. phere, biosphere, hydrosphere, cryosphere, and lithosphere (Figure 3.9). The climate system is coupled, in the sense that the There remains much uncertainty as to how El Niño/La Niña components interact at many spatial and temporal scales. conditions may change with a warmer climate. While some studies suggest that with increased greenhouse gas concentrations, the El The Earth’s climate is determined by the long-term balance Niño pattern may become more dominant, there is great uncer- between incoming solar radiation and outgoing terrestrial radia- tainty surrounding this issue (Collins, 2005 ). tion. Incoming solar radiation is partly absorbed, partly scattered, and partly refl ected by gases in the atmosphere, by aerosols, and For the NAO, some scientifi c evidence including modeling by clouds. Under equilibrium conditions, there is an energy bal- studies suggests that, with a warmer atmosphere, ocean tempera- ance between the outgoing terrestrial or “longwave” radiation tures will rise in a pattern that reinforces the positive phase of and the incoming solar or “shortwave” radiation. Greenhouse the NAO. As was the case with projections of ENSO with global gases are responsible for an approximately 30 °C elevation of warming, future patterns of the NAO also remain extremely global average surface temperature. Since the Industrial Revolu- uncertain (Visbeck et al. , 2001 ). tion, increasing greenhouse gas concentrations due to fossil fuel combustion, cement-making, and land use changes has increased the mean surface temperature of the Earth by approximately an 3.3 Global climate change and its impact additional 1 °C. on urban areas These and other climate changes and impacts have been documented by an international panel of leading climate scien- Along with all other planetary surfaces, urban climates are tists, the Intergovernmental Panel on Climate Change (IPCC), subject to global changes due to radiative forcing. The impact of formed in 1988 to provide objective and up-to-date information global change on cities is the subject of this section. regarding the changing climate. Key fi ndings of the 2007 Fourth

Figure 3.9: The global climate system. Shown are the many interactions among the different components of the climate system, which includes cities. Source: IPCC, WGI (2007). 54 Urban climate: Processes, trends, and projections

Assessment Report (AR4; IPCC, 2007) included the following Carbon dioxide, which accounts for over 75 percent of all green- (as summarized in Horton et al ., 2010): house gas emissions, is emitted into the atmosphere through the • there is a greater than 90 percent chance that warming tem- burning of fossil fuels and the clearing of land for agriculture . peratures are primarily due to human activities (IPCC, 2007); Methane is emitted from natural gas production, livestock, and • atmospheric concentrations of carbon dioxide (CO2) are now agricultural production. Application of nitrogen fertilizer for more than one-third higher than pre-industrial levels; agricultural production also leads to nitrous oxide being emitted • concentrations of other important greenhouse gases methane into the atmosphere. Most hydrofl uorocarbons, another impor- (CH4) and nitrous oxide (N2O) have increased by more than tant greenhouse gas, come from industrial activity. 100 percent and close to 20 percent respectively over the same time period; 3.3.2.2 Aerosols • further increases in greenhouse gas concentrations are projected to lead to further temperature increases and associated Aerosols are atmospheric particles of both natural and anthro- changes in the climate system; pogenic origin. Natural aerosol sources include volcanoes and • over the twenty-fi rst century, global average temperature is sea salt, while key anthropogenic sources include fossil fuel expected to increase by between 1.8 and 4.0 °C. combustion and biomass burning. Aerosol concentrations tend to be higher in urban than rural areas, although during times Warming is expected to be largest over land and in the high- of extensive biomass burning rural areas can have comparable latitude North, where some Arctic cities may experience warming concentrations. Climatic, human health, and visibility effects of exceeding 8 °C by 2100. Outside the tropics and subtropics, aerosols have been documented. (Shu et al., 2000 , 2001 ; Pawan the largest warming is generally expected in winter. Generally et al ., 2006 ; Eri et al ., 2009 ). Aerosols modify the earth’s energy speaking, precipitation is expected to increase in high-latitude budget by scattering and absorbing short- and longwave radia- cities and decrease in subtropical cities. Hot extremes and cold tion. As a prime radiative forcer (Charlson et al. , 1992 ), aerosol extremes in cities are generally expected to increase and decrease particles also infl uence cloud optical properties, cloud water respectively. content, and lifetime. That is referred to as the indirect effect of aerosols, or indirect climate forcing (Harshvardhan, 2002 ). Volcanic aerosols can lead to brief periods of global cooling, and As CO2 continues to be absorbed by the oceans, ocean acidifi cation will accelerate, with potentially large implications on biomass burning of agricultural regions has been shown to affect marine ecosystems. While the implications may be largest for regional weather. While greenhouse gas effects are primarily those coastal cities where marine ecosystems are a source of global and regional in scale, aerosol effects span from the global economic livelihood and sustenance, in a global economy all the to the urban scale . world’s cities would be indirectly impacted by large-scale ocean acidifi cation. 3.3.2.3 Land use change

Land use change and urbanization infl uence the climate 3.3.2 Drivers of global climate change through changes in surface albedo , land roughness, hydrological and thermal features. Across the globe, human activities have The global changes described above can largely be attrib- changed the face of the planet. Deforestation has modifi ed the uted to three drivers, or causes, of climate change: greenhouse climate by changing solar absorption and moisture transfer gases, aerosols, and land use changes. These three drivers, with rates, as well as increasing carbon dioxide levels. Agricultural emphasis on urban contributions, are described below. expansion has modifi ed these processes as well. Urbanization and development have led to decreased groundwater absorption 3.3.2.1 Greenhouse gases by the land and more heat absorption as the built environment expands (Zheng et al ., 2002 ; Gao et al ., 2003 ; Pielke, 2005 ; Lian Gases that trap heat in the atmosphere are referred to as green- and Shu, 2007 ). Like aerosol effects, the climatic effects of land house gases. Three primary greenhouse gases that are directly use changes range from large regional scales to urban scales. linked to anthropogenic activities are carbon dioxide (CO2 ) , methane (CH4 ), and nitrous oxide (N 2 O). As centers of population , economic activity, and energy use , cities are responsible 3.4 Observed climate change in cities for a large portion of greenhouse gas emissions. On a per capita basis, however, urban residents in developed nations probably have lower emission rates than ex-urban residents due to the We present observed long-term climate data for the 12 cities to inherent energy effi ciencies of mass transit and multiple-resident evaluate whether there are long-term trends. The urban heat island buildings. effect that may exist in each case should in principle be captured by the urban weather stations used. The selected cities include Some greenhouse gases, such as carbon dioxide , occur in the a wide range of climate zones. Tropical cities such as Dakar are atmosphere through natural processes, while others, such as the characterized by warm temperatures throughout the year. The mid- hydrofl uorocarbons, are created solely from human activities. latitude cities, in contrast, all experience a continental climate, 55 Climate change and cities

with large temperature differences between summer and winter Table 3.2: Observed climate trends in cities. seasons. Some cities such as Delhi have a monsoon al climate , with Temperature the majority of the precipitation occurring during one part of the Number of trend ( °C per year, while other cities such as New York experience substantial City Years missing months decade) precipitation during all months. These differences in background climate can be thought of as the foundation for the unique blend of Athens 1900–2005 15 0.03 climate change hazards each city faces, since climate changes are Dakar 1900–2005 57 –0.09** superimposed on a city’s baseline climate. Delhi 1931–2005 10 0.03 Over the past century, there have been signifi cant observed Harare 1900–2002 44 –0.06** trends in climate hazards including annual mean temperature, Kingston 1943–2005 59 0.19** annual precipitation, and extreme events , such as heat waves and intense precipitation, at global scales (IPCC, 2007 ). Trend London 1900–2005 0 0.08** analyses are a fi rst step towards attributing changes to factors Melbourne 1900–2005 0 0.14** ranging from variability to greenhouse gases, to changes in the urban environment. New York 1900–2005 0 0.16** São Paulo 1900–2005 30 0.27** The observed annual temperature and precipitation values for Shanghai 1900–2005 33 0.05** each city were computed using a monthly data set.1 The months in the data set that were missing were replaced by the climatolog- Tokyo 1900–2005 2 0.26** ical average for that month over the full time series for the city. Toronto 1938–2005 4 0.10 The period over which the observed trends were analyzed was 1900 to 2005 unless otherwise noted. For the temperature trends, Number of Precipitation trend the following cities did not have data sets running the entire City Years missing months (mm per decade) length of this period; Toronto (1938–2005), Delhi (1931–2005), Athens 1900–2005 11 2 Kingston (1943–2005), and Harare (1900–2002). For the pre- Dakar 1900–2005 24 –20** cipitation trends, Toronto (1938–2005) and Harare (1937–2002). A brief city-by-city summary for observed trends in temperature Delhi 1900–2005 18 16* and precipitation is shown in Table 3.2 . The statistical signifi - Harare 1937–2002 26 –21 cance of the trends is also included. Graphs of the observed climate data for each of the 12 cities are shown in Figure 3.10 . Kingston 1900–2005 19 –15 London 1900–2005 0 2

3.4.1 Temperature Melbourne 1900–2005 0 –4

Observations of annual mean temperature are used to deter- New York 1900–2005 0 18 mine whether or not a city is experiencing warming or cooling. São Paulo 1900–2005 16 29** In addition to mean temperature, maximum temperature and Shanghai 1900–2005 24 –3 minimum temperature can also be used to identify changes in the climate. For these variables, fi ner temporal scales (daily data) are Tokyo 1900–2005 2 –17* useful for analysis of observed climate trends . However, because Toronto 1938–2005 4 7 of the limited data records for some cities, obtaining climate data on these small timescales can be diffi cult. These additional tem- Annual temperature and precipitation statistics are computed for all cities using data from the National Climatic Data Center Global Historical Climatology Network (NCDC perature variables can also be used in the analysis of extreme GHCN v2 ), UK Met Offi ce and Hadley Centre, Australian Bureau of Meteorology, and climate events, including hot days and heat waves. Environment Canada. The single star (*) indicates that the trend is signifi cant at the 95 percent level and the double star (**) indicates that trend is signifi cant at the 99 percent level. For the 12 cities that were selected for the chapter, observed trends in annual mean temperature showed warming in 10 of the cities. Over the past century the most rapid rate of warming Although most of the cities did show a warming trend, the has occurred in São Paulo. The city had a trend of +0.27 °C trends in each individual city vary by their rate of change. For per decade. The two cities that showed cooling (temperatures example, one city may have had a slow increase for a large part decreasing over the time period) were in Africa . Dakar and Harare of the twentieth century with a rapid warm up at the end of the had observed trends in annual mean temperature of –0.09 °C and period, while another may have seen a moderate increase over –0.06 °C respectively. However, for these two cities, while the the entire time period. Further investigation of the trend for each overall trend is negative, both have seen warming in the past two individual city is necessary to understand the possible causes decades. and potential impacts of warming temperatures. The rate at

1 For data sources, see Table 3.2 56 Urban climate: Processes, trends, and projections which the urban heat island is increasing in each city will affect 3.4.1.1 Tokyo’s heat island observed temperature trends. For example, approximately one- third of the warming trend in New York City has been attributed As temperatures rise in Tokyo , residents are experiencing more to urban heat island intensifi cation (Gaffi n et al., 2008 ). In devel- health problems, including heatstroke and sleeping diffi culties. Both oping countries, the rate is probably higher (Ren et al. , 2007 ). are associated with higher nighttime temperatures (Figure 3.11).

Figure 3.10: Observed climate trends in cities. Trends and statistical signiﬁ cance are shown for the data available for the twentieth century (see Table 3.2 for speciﬁ c years for each city). Note differences in temperature and precipitation scales. 57 Climate change and cities

Figure 3.10: (continued)

Additional impacts include ecosystem changes such as earlier ﬂ ow- the period from 1900 to 2005, mean temperature in the city ering dates and changes in insect and amphibian populations. rose at a rate of 0.27 °C per decade. Much of the warming is occurring at night, with minimum temperature increasing at a 3.4.1.2 Observed temperature trend in São Paulo, Brazil faster rate than maximum temperature. The average of these two variables, maximum and minimum temperatures, is the For the cities chosen for the chapter, São Paulo had the mean temperature. São Paulo’s warming was greater in winter most rapid rate of increase in mean annual temperature. Over than in summer. Some of the warming in São Paulo may be 58 Urban climate: Processes, trends, and projections

Figure 3.10: (continued) caused by land use changes due to urbanization (Marengo and slices can be explained by changes in the frequency of modes of Camargo, 2008 ). natural climate variability . ENSO events in recent years coin- cide with the more rapid rates of warming. In addition, warming The observed warming in mean annual temperature appears temperatures in the South Atlantic Ocean also may inﬂ uence to be greatest over approximately the past 20 years, as compared the warming mean temperatures in São Paulo (Marengo and to the 20 years prior to that. Some of this variation between time Camargo, 2008 ) . 59 Climate change and cities

Figure 3.11: Urban heat island impacts in Tokyo, Japan. The number of people sent to the hospital with heatstroke has been increasing over time. Source: Tokyo Metropolitan Government

3.4.2 Precipitation

Observations of annual total precipitation reveal precipitation changes in cities on decadal timescales. Natural climate variability can greatly infl uence annual precipitation (more so than temperature). Changes in the inter-annual variability of precipitation can also provide insight into how the climate is changing. For example, in New York City, while annual precipitation has only increased slightly in the past century, inter-annual variability has become more pronounced. In addition, analyzing trends in monthly rainfall can also be useful, as many cities experience two distinct seasons, such as the monsoons in Delhi, India . Data on shorter timescales, such as daily and weekly, can be used in extreme events analysis of droughts, fl oods, and intense precipitation rates. Again, with limited data records for some cities, obtaining climate data on these small timescales can be diffi cult.

Precipitation trends for the 12 cities refl ect the regional nature of precipitation changes with climate change, relative to the largely homogenous changes expected in temperature. Of the 12 cities looked at in the chapter, half saw increasing trends in Figure 3.12: Observed precipitation trends from 1938 to 2005 in Toronto, Canada: annual precipitation. Over the past century the most rapid rate rain above; snow below. Showfall trend is signifi cant at the 95% level. of precipitation increase has occurred in São Paulo (+29 mm Source: Data are from Environment Canada. per decade), and the largest decrease has occurred in Harare (–21 mm per decade). inter-annual variability of precipitation. Cities with increasing For the cities where observed trends in annual precipitation annual precipitation over the time period still may have pro- showed increasing (decreasing) precipitation, the trends do nounced periods of drought, while those experiencing a drying not reveal whether the wetter (drier) conditions are caused by trend may still have years with higher than average rainfall. more (less) frequent heavy rainfall events or more (less) per- sistent lighter rainfall. Also, the direction of trends for extreme 3.4.2.1 Observed precipitation trend in Toronto, Canada precipitation events, such as days where precipitation is greater than 50 mm, will not necessarily correspond to the trend in Toronto is one city selected for this chapter that receives pre- annual precipitation. For example, a city could experience a cipitation in the form of snow. While total precipitation has been decreasing trend in annual precipitation, but have an increase in increasing over the past century in the city, most of the increase short-duration intense precipitation events. has been in the form of more rainfall. Snowfall in Toronto has been on the decline, leading to a decrease in the percentage of total pre- As was the case with temperature, there are multidecadal cipitation falling as snow. The trends for observed annual rainfall fl uctuations in observed precipitation trends specifi c to each city. and snowfall in Toronto are presented in Figure 3.12 , with rain- This is especially true for this climate variable given the large fall and snowfall shown in green and blue respectively. Rainfall 60 Urban climate: Processes, trends, and projections has increased at a rate of 10 mm per decade while snowfall has included increases in snowfall, especially during the winter decreased at a rate of 4.3 cm per decade between 1938 and 2008. months. Temperatures in these areas of the country have warmed enough to allow for more precipitation but have not crossed the To some extent, trends in precipitation for a given city depend threshold for snow to become rain (Zhang et al. , 2000 ) . on annual mean temperatures. In a warmer climate, the atmosphere can hold more moisture, allowing for more precipitation. 3.4.3 Sea level rise But higher temperatures also can alter the type of precipitation that will fall. For Toronto, warmer temperatures have allowed Sea level rise analysis was not performed for all of the for more precipitation; in the spring, precipitation that once fell coastal cities due to limited data sets. However, coastal cities are as snow is now falling as rain. This trend is not evident in many extremely vulnerable to rising sea levels, since approximately of the colder parts of Canada. In colder areas the increases in 35 percent of world population lives within 100 km of the coast total annual precipitation (like those observed in Toronto) have (Hachadoorian et al., 2011 ; L. Hachadoorian, pers. comm.).

[ADAPTATION] Box 3.2 Sorsogon City, Philippines, responding to climate change

Bernhard Barth and Laids Mamonong devastation within a two-month interval and leaving in their wake a total of 27,101 families affected and 10,070 totally UN-HABITAT damaged houses (Box Figure 3.2). The fi rst typhoon, in just Sorsogon City is one of 120 cities in the Philippines in the 5 hours, caused damage to public infrastructure estimated at Asia Pacifi c region. It has a land area of 313 square kilom- 208 million pesos or US$4.3 million. The city is projected to eters with a population of 151,454 (as of 2007) growing at a experience more cases of prolonged monsoon rains resulting rate of 1.78 percent annually. Its economy is based mainly in total rainfall exceeding 2,800 to 3,500 mm per year. on agriculture , fi shing, trade, and services. It is the capital The V&A assessment revealed that the city’s geographical and the administrative, commercial, and educational center location and previous stresses make it sensitive to changes of Sorsogon Province. in extremes – such as tropical cyclones , storm surges, and In August 2008 the city launched a Climate Change Initiative, extreme rainfall/fl ooding – and changes in means – such as championed by the new mayor. Until then the popular perception increased temperature, increased precipitation, and sea level was that climate change is a global and national issue requiring rise. With sea level rise projected to accelerate, the city’s built- limited action from the local government. A series of briefi ngs for up areas situated near the coast present the highest vulnera- decision- makers and local leaders was conducted to enhance bility to climate change impacts because they have the highest basic understanding of climate change and the important role concentration of people, especially informal settlers, living in of local government. This resulted in an expressed commitment inadequate structures in danger zones. These areas are also from decision-makers in developing their city’s climate change the hub for economic activities (accounting for 60 percent of profi le and defi ning responsive local actions. the economy) and the location of basic lifelines such as water, electricity , and basic service facilities. Around 36 percent of the Various city stakeholders worked together with the local govern- total population, or 55,000 people, are vulnerable to fl ooding. ment in the conduct of a participatory vulnerability and adapta- Over 35,000 people from nine coastal villages are threatened tion (V&A) assessment . Using climate change projections and by sea level rise and storm surge, of whom 22,000 are women. risk assessments from national government agencies and private research institutions, the city government developed its local vulnerability assumptions. To assess local impacts, the city gathered and analyzed its own recorded observations. These were further substantiated by local people’s accounts of their personal experiences. During city consultations, residents recounted how typhoons and storm surges over the past decade had become stronger and more destructive. These records and personal accounts were recorded as evidence of climate change impacts through community risk mapping. Using hand- drawn maps, local people graphically described the changes in the reach of tidal fl ooding and identifi ed the areas gradually lost due to sea level rise and erosion. This participatory exercise promoted ownership by the locals of the assessment process and results, and increased their awareness of climate change impacts. Moreover, the process empowered the people to work together with the local government in fi nding practical solutions Box Figure 3.2: Section of seawall in Beacon District, Sorgoson, partially that they can personally act on. destroyed by a 2006 storm. Many of Sorgoson’s informal settlements are just As noted in the city’s climate change profi le, the city was behind the seawall. badly hit by two super-typhoons in 2006, causing widespread Source: UN-HABITAT/Bemhard Barth

61 Climate change and cities

Knowing these climate change vulnerabilities (areas, popula- associations for convergence of efforts on adaptation and tion, economic activities, policy gaps), the city government mitigation. is now engaging local communities and the private sector in 2. The city government’s capacity must be developed to climate change adaptation planning. Using tools from UN- make it more responsive to increase its resilience to cli- HABITAT ’s Sustainable Cities Programme, the local govern- mate change impacts. A framework must be developed ment conducted multisector city consultations that resulted to help and guide the city in integrating climate change in the identiﬁ cation of four priority “quick-win” responses to considerations in the land use and development plans. increase people’s resilience to climate change, namely: (i) A stronger link with national climate change programs is improving settlements and basic infrastructure, (ii) enhancing critical especially in enhancing building code and land use livelihoods, (iii) developing climate and disaster risk manage- planning parameters. ment systems, and (iv) improving environmental manage- 3. The city needs to learn from good practices by other cities. ment and climate change mitigation actions. Issue working It should also share its own experience in engaging various groups composed of representatives from people’s organi- stakeholders in deﬁ ning a collective climate change action. zations, NGOs, private sector, and LGU were organized to 4. It is crucial for the business sector to play a vital role in develop the action agenda per “quick-win” area and ensure providing green building technology development and in its implementation. promoting risk-resilient communities through the use of appropriate and innovative technologies in housing and So far the following important lessons have been learned: infrastructure development. 1. There is a need to promote and advocate awareness on climate change among the general public and stake- The above lessons have become major considerations as the holders through various media and community activities. city works on mainstreaming climate change risk manage- This would broaden/establish partnerships among the pri- ment into its local governance processes and implementing vate, public, academic, civil society, and neighbourhood local climate change adaptation actions . Source: UN-HABITAT (2010). Sorsogon City Climate Change Vulnerability and Adaptation Assessment. Available at: http:www.unhabitat.org.ph/climate-change/ knowledge-productsoulputs.

Cities in this chapter at risk of sea level rise include New York, Limited data availability at short timescales constrains anal- São Paulo , Tokyo , London , Shanghai , Melbourne , and Dakar . ysis of urban extreme events. Extreme events also differ by city; For all of these coastal cities , the sea level rise they are expe- for many inland locations, for example, coastal storm surges can riencing is caused by a combination of global and local fac- be ignored. Some extreme events occur more frequently in asso- tors. While the rate of sea level rise for each city due to global ciation with certain phases of climate variability patterns such thermal expansion and meltwater from glaciers and ice sheets is as ENSO. Because some variability patterns are somewhat pre- the same, city-speciﬁ c factors include land subsidence and local dictable, there is an opportunity for seasonally forecasting these ocean height. These city-speciﬁ c terms are necessary not only to events, allowing cities to prepare for them in advance . determine the local rate of sea level and compare it to other cities and the global trend, but also for sea level projections . 3.4.4.1 Hot days in Melbourne, Australia

3.4.3.1 Observed sea level rise in Dakar, Senegal One city that experiences hot days and heat waves is Mel- bourne . As defi ned by the World Meteorological Organization One city that is at risk of and has been experiencing rising (WMO), hot days have maximum temperature exceeding 35 °C. sea levels over the past century is Dakar . Unfortunately, due to For Melbourne, the trend from 1900 to 2008 shows no signifi - limited data availability, a common phenomenon in developing cant increase in the number of hot days, shown in Figure 3.13. country cities, very little can be said about sea level rise trends. The trend in hot days does not reveal a signifi cant increase, even While sea level has been increasing at a rate of 1.5 cm per decade though annual average temperatures have risen signifi cantly over over the 11-year data record, for such short timescales natural the past century. variations and cycles can dominate any climate change signal. This example highlights the need for expanded data collection Combined with prolonged periods of dry weather, consecu- and quality control in many cities. tive hot days (heat waves) have the potential to greatly impact cities and their surrounding areas. Specifi cally for Melbourne, 3.4.4 Extreme events years with above-average hot days, combined with other meteorological conditions, yield an increased threat of wild fi res . Extreme events can be defi ned as climate variables experi- Although the fi res themselves are often in outlying areas away enced in a limited duration. Temperature extremes include hot from the city, infrastructure, agriculture , ecosystems , water, and days where temperatures exceed a specifi ed threshold, and heat human resources critical to the city’s survival may be impacted. waves – consecutive hot days. Precipitation extremes, which Fires can also directly impact the city by reducing air quality. cover varying timescales, include intense/heavy precipitation The most recent extreme warmth in early 2009, the heat wave of events and droughts. Coastal storms and tropical cyclones are 2006, and in the summer of 1983 are all examples of years with also types of climate extreme events. increased fi re activity and high numbers of hot days . 62 Urban climate: Processes, trends, and projections

Figure 3.13: Observed temperature extremes, hot days with temperatures above 35 °C, in Melbourne, Australia. Source: Data are from Bureau of Meteorology, Australia.

3.4.4.2 Drought in Harare, Zimbabwe fi rst step for developing adaptation strategies. The projection component is derived from the global climate models that focus Drought is a precipitation extreme event that occurs over on broad geographical scales, rather than urban climate processes. longer timescales, ranging from months to years. Unlike other However, historical data are used for projection baselines, and extreme events, droughts lack a formal defi nition or index that is these baselines refl ect the effects of urban climate processes. applicable globally, which makes assessments of their severity, Some of the challenges of and potential solutions to incorpo- trends in their frequency, and future projections diffi cult. Because rating urban processes in city-specifi c climate projections are of the varying indices and defi nitions, a qualitative assessment discussed throughout the section. of drought based on precipitation is appropriate for multi-city analyses. 3.5.1 Global climate change projection methods

One city that experiences frequent and prolonged periods of 3.5.1.1 Global climate models drought is Harare, Zimbabwe. Over the past century, droughts have occurred several times, including 1991/1992, 1994/1995, The theoretical impact of atmospheric greenhouse gases on the and 1997/1998. Analysis of precipitation data reveals that pre- planet’s energy budget has been recognized since the nineteenth cipitation in Harare has been declining over the past century at century (Arrhenius, 1896). Global climate models (GCMs) are a rate of –21 mm per decade. In addition, of the ten driest July– mathematical representations of climate system interactions that June periods between 1938 and 2002, fi ve have been since 1980. help quantify the impact of greenhouse gases and other system These results suggest that drier conditions may be becoming interactions. Climate models also include other climate drivers, more frequent. including aerosols, land-cover changes, and solar variability. Figure 3.14 depicts physical processes that a climate model sim- The droughts that occur in Harare have a strong connection ulates. to ENSO events. While ENSO is not the only factor that affects droughts in Zimbabwe, using this as one forecast predictor can The simulations conducted for the IPCC 2007 report were help the city prepare and issue drought warnings with ample lead- run at higher spatial resolution than prior simulations and time. However, reliance on just this predictor can be dangerous, as more accurately included complex physical processes such as was the case in 1997/1998. Substantial preparations for a drought turbulent fl uxes. Current-generation climate models capture were made that year with the prediction of a strong El Niño, yet many aspects of climate, including twentieth-century warming conditions did not become dry as expected (Dilley, 2000 ). (when simulated using historical greenhouse gas concentrations and volcanic aerosols) (Figure 3.15). They also capture 3.5 Future climate projections key climate characteristics of paleoclimates such as the cool Last Glacial Maximum (~21,000 years ago) and warm mid- Holocene (~6,000 years ago). Model successes over a range This section describes climate change projection methods of historical climate conditions increase confi dence that future and results. The projections can be used by urban stakeholders simulations will be realistic as CO2 concentrations continue to identify key sector-specifi c impacts of climate change, as a to increase. 63 Climate change and cities

Figure 3.14: Global Climate Model. Global climate models break the surface down into a series of gridboxes. Within each gridbox, equations are used to simulate the physical processes in the climate system. Source: NOAA.

3.5.1.2 Sources of uncertainty: emissions scenarios and • Sensitivity of the climate system to changing concentrations climate sensitivity of climate drivers; • Climate variability, which is largely unpredictable beyond Despite climate model advances, projections continue to be one year in advance, and can mask the climate change signal characterized by large uncertainty (for more information, see at urban scales; Horton et al., 2010). The critical uncertainties concern: • Local physical processes such as the urban heat island, and • Future concentrations of greenhouse gases and other climate coastal breezes that occur at smaller spatial scales than GCMs drivers that alter the global energy balance including aerosols can resolve. and black carbon; These uncertainties are partially addressed here by using a suite of GCMs and greenhouse gas emissions scenarios, aver- aged over 30-year time periods to cancel out a portion of the unpredictable natural variability. By presenting the projections as changes through time, uncertainties associated with local physical processes are minimized, although they cannot be elim- inated entirely.

3.5.1.3 Greenhouse gas emissions scenarios

Greenhouse gas emissions scenarios, described in the IPCC Special Report on Emissions Scenarios (SRES; Nakicenovic et al., 2000), are plausible descriptors of possible future socio- economic, technological, and governance conditions that drive Figure 3.15: Twentieth-century observations and GCM results. The observed temperature energy demand and fuel choices (Parson et al., 2007). The green- series follows the model simulation that includes both human effects and natural forces. house gas concentrations associated with each scenario were used Source: United States Global Change Research Program (2009). as inputs for global climate model simulations. Three greenhouse 64 Urban climate: Processes, trends, and projections

Table 3.3: IPCC AR4 climate models.

Atmospheric resolution (latitude × Institution Model longitude)

Bjerknes Centre for Climate BCCR 2.8 × 2.8 Research, Norway

National Weather Research CNRM 2.8 × 2.8 Centre, METEO-FRANCE, France

Canadian Center for Climate CCCMA 3.75 × 3.75 Modeling and Analysis, Canada

CSIRO Atmospheric CSIRO 1.9 × 1.9 Research, Australia

Geophysical Fluid Dynamics GFDL1 2.0 × 2.5 Laboratory, USA Figure 3.16: Atmospheric CO2 concentrations resulting from three emissions Geophysical Fluid Dynamics GFDL2 2.0 x 2.5 scenarios. Observed CO2 concentrations through 2003, and future CO2 for 2004 to Laboratory, USA 2100. The table gives the value of the CO2 concentration for each scenario in 2050 and 2100. NASA Goddard Institute for GISS 4.0 × 5.0 Source: IPCC (2007). Space Studies

Institute for Numerical INMCM 4.0 × 5.0 Mathematics, Russia gas emissions scenarios (A2, A1B, and B1), were selected for use in this chapter. Figure 3.16 shows the CO2 concentration asso- Pierre Simon Laplace IPSL 2.5 × 3.75 ciated with each of these three SRES scenarios. The three sce- Institute, France narios span a reasonably wide range of concentrations and allow Frontier Research Center for MIROC 2.8 × 2.8 for nominal high-, medium- and low-path CO2 futures. Global Change, Japan Meteorological Institute MIUB 3.75 × 3.75 Greenhouse gas concentrations could exceed the levels in of the University of Bonn, these SRES scenarios, due to increases in human emissions (for Germany example, more rapid economic and population growth) or due to carbon and methane cycle feedbacks. As an example of the latter, Max Planck Institute for MPI 1.9 × 1.9 Meteorology, Germany increasing Arctic temperatures are leading to accelerated perma- frost melting, which could lead to the release of stored methane Meteorological Research MRI 2.8 × 2.8 to the atmosphere. As greenhouse gas emissions and concentra- Institute, Japan tions are monitored in the coming decades, it will be possible to National Center for CCSM 1.4 × 1.4 better assess the likelihood of high-end emissions scenarios. Atmospheric Research, USA

National Center for PCM 2.8 × 2.8 3.5.1.4 Global climate model selection Atmospheric Research, USA

From the IPCC climate model database, we selected 16 Hadley Centre for Climate HadCM3 2.5 × 3.75 prominent models that had available output for each of the three Prediction, Met Ofﬁ ce, UK emissions scenarios. The selected models and their space resolution are listed in Table 3.3. The outputs are collected by the World Climate Research Program (WCRP) and the Program 3.5.1.5 Model-based probability for Climate Model Diagnosis and Intercomparison (PCMDI) (www-pcmdi.llnl.gov/ipcc/about_ ipcc.php), at the Lawrence- Our combination of 16 global climate models and three emis- Livermore Laboratory in Livermore, California. sions scenarios produces a distribution of 48 outputs for temperature and precipitation – in essence a “model-based” probability While it is unlikely that the future will closely follow a single distribution. As described in Horton et al. (2010), future time emissions scenario, or that a GCM projection will capture all periods are compared to a 1971–2000 model baseline. Temper- aspects of the climate system’s response, using a suite of emis- ature change is expressed as a difference, and precipitation as sions scenarios and climate model simulations increases the a percentage change. It should be noted that the model-based probability that the projection range will span the actual future probability distribution will differ from the true probability climate outcome. More information about the rationale for this distribution, for a range of reasons described in Horton and approach, used by the IPCC, can be found in IPCC, 2007. Rosenzweig (2010). 65 Climate change and cities

3.5.1.6 Downscaling 3.5.2.1 Temperature

As described in Horton and Rosenzweig (2010), the projec- All cities are projected to see warming, with temperature tions use GCM output from a single model gridbox. Depending increases of between 1.5 and 5.0 °C possible by the 2080s. The on the model, resolution can be as ﬁ ne as ~75 – ~100 miles city expected to see the greatest warming is Toronto , Canada , or as coarse as ~250 – ~275 miles. Changes from the relevant with projected increase in temperature by the 2080s ranging gridbox are applied to observed station data, a procedure known from 2.5 to 5.0 °C. as the Delta Method that corrects for biases over the baseline period caused by factors such as the spatial mismatch between a Warming is expected to increase with distance from the gridbox and a point location. More sophisticated statistical and equator; Toronto , for example, is expected to warm approxi- dynamical downscaling techniques can apply more localized mately two times as much as Kingston by 2100 under the A2 change projections. While such techniques should be pursued in scenario. Warming is also generally expected to be greater in the future, a host of issues may hinder their utlility. These issues interior continental regions than in coastal regions , due to the include: the computational cost of dynamical downscaling and moderating inertial inﬂ uence of the ocean s. This helps explain the fact that statistical downscaling may be less valid as climate why more warming is expected in Toronto than London , even statistics change (Christensen et al., 2007). though they are at comparable latitudes. In the extra-tropical regions, warming will generally be greatest in winter. 3.5.2 Temperature, precipitation, sea level rise, and extreme events climate projections Temperature projections for Delhi, India

The climate projections for temperature and precipitation for Delhi, India, is one city where annual mean temperature is pro- the 12 cities are presented fi rst. For sea level rise and extreme jected to increase over the next century. A slight upward, although event projections, more detailed methodologies are provided in not statistically signifi cant, trend has been observed in the city from a case study format focusing on New York City. 1931 to 2005. This trend is consistent with trends from the region around Delhi in India. The region as a whole has experienced a It is important to recognize the uncertainties associated with warming trend over the past century, with the greatest warming climate models . Therefore, the direction of change suggested here occurring in the past three decades. The long-term warming trend is as important as the specifi c numbers, which are not precise. appears to be driven by warmer maximum temperatures, while the Table 3.4 summarizes the temperature and precipitation projec- more recent trend is a result of warmer maximum and minimum tions for each of the cities for three future timeslices. Shown is temperatures (Kothawale and Rupa Kumar, 2005 ). the middle 67 percent (central range) of the model projections. Graphs showing the projections for each of the cities are shown Looking at Table 3.4, Delhi is projected to see an increase in Figure 3.17 . in annual mean temperature of between 2.5 and 4.5 °C by the

Table 3.4: GCM projections for temperature and precipitation.

Temperaturea Precipitationa

City Name 2020s 2050s 2080s 2020s 2050s 2080s

Athens 0.5 to 1.5 °C 1.0 to 4.0 °C 2.0 to 4.0 °C –10 to +0% –20 to –5% –25 to –10%

Dakar 0.5 to 1.0 °C 1.0 to 2.0 °C 1.5 to 3.0 °C –10 to +10% –15 to +10% –20 to +15%

Delhi 0.5 to 1.5 °C 1.5 to 2.5 °C 2.5 to 4.5 °C –10 to +20% –15 to +35% –15 to +35%

Harare 0.5 to 1.5 °C 1.5 to 2.5 °C 2.0 to 4.0 °C –10 to +5% –10 to +5% –15 to +5%

Kingston 0.5 to 1.0 °C 1.0 to 1.5 °C 1.5 to 2.5 °C –10 to +5% –25 to +0% –30 to –5%

London 0.5 to 1.0 °C 1.0 to 2.0 °C 1.5 to 3.0 °C –5 to +5% –5 to +5% –5 to +5%

Melbourne 0.5 to 1.0 °C 1.0 to 2.0 °C 1.5 to 3.0 °C –10 to +5% –15 to +0% –20 to +0%

New York 1.0 to 1.5 °C 1.5 to 3.0 °C 2.0 to 4.0 °C +0 to +5 % +0 to +10% +5 to +10%

Sao Paulo 0.5 to 1.0 °C 1.0 to 2.0 °C 1.5 to 3.5 °C –5 to +5% –5 to +5% –10 to +5%

Shanghai 0.5 to 1.0 °C 1.5 to 2.5 °C 2.0 to 4.0 °C –5 to +5% +0 to +10% +0 to +15%

Tokyo 0.5 to 1.0 °C 1.5 to 2.5 °C 2.0 to 4.0 °C –5 to +5% +0 to +5% +0 to +10%

Toronto 1.0 to 1.5 °C 2.0 to 3.0 °C 2.5 to 5.0 °C +0 to +5% +0 to +10% +0 to +15%

Shown is the central range (middle 67 percent) of values from model-based probabilities; temperatures ranges are rounded to the nearest half-degree and precipitation to the nearest 5 percent. a Based on 16 GCMs and three emissions scenarios. 66 Figure 3.17: Observed and projected temperature and precipitation. Combined observed (black line) and projected temperature and precipitation is shown. Projected model changes through time are applied to the observed historical data. The three thick lines (green, red, and blue) show the average for each emissions scenario across the 16 GCMs. Shading shows the central range. The bottom and top lines, respectively, show each year’s minimum and maximum projections across the suite of simulations. A 10-year ﬁ lter has been applied to the observed data and model output. The dotted area between represents the period that is not covered due to the smoothing procedure. WCRP, PCMDI, and observed data sources found in Table 3. 2. Note differences in temperature and precipitation scales. Figure 3.17: (continued) 68 Figure 3.17: (continued) 69 Climate change and cities

2080s. Figure 3.17 shows the full range of GCM projections for As shown in Table 3.4 Athens is projected to see drops in annual mean temperature for Delhi. Only around the 2030s do precipitation between 10 and 25 percent by the 2080s. Although differences between the emissions scenarios emerge . seasonal projections are less certain than annual results, the climate models project much of the drying to occur during the 3.5.2.2 Precipitation summer months. Figure 3.17 shows the full range of GCM projections for annual total precipitation in Athens. Even using There is greater variability in the direction of projections of 10-year smoothing, there remains large historical variability in percentage change in precipitation for these cities. Some cities the observed trend. For the projections, only from approximately are expected to see increases in precipitation while others are the 2040s onward does the B1 scenario produce smaller precipi- projected to experience sharp declines in precipitation by the tation decreases compared to the other two scenarios . 2080s. For cities in higher latitudes projected to see increased precipitation, such as Toronto and New York, most of the pre- 3.5.2.3 Sea level rise cipitation increase will be in the form of rainfall, as snow fall is likely to decline given the warmer temperatures. As the ocean s warm and expand and land-based ice continues to melt, all the coastal cities analyzed here are expected to expe- Precipitation changes will differ dramatically by region. In rience sea level rise this century. However, the rate will differ general, mid-to-high latitude cities such as Toronto, New York, by city, for two primary reasons. First, the local height of the and Tokyo are expected to experience precipitation increases. adjacent ocean can differ by city, due to the infl uence of ocean However, some mid-latitude cities, such as London , are expected currents , water temperature, and salinity, and the infl uence of to experience signifi cant summer-time drying, which could lead wind and air pressure. Second, local land height change can to net precipitation decreases. Other cities at the boundaries differ by city. Some cities such as Shanghai are sinking due to between the mid-latitudes and subtropics are expected to expe- the effects of groundwater extraction and compaction of soils by rience drying, including Harare and Melbourne . Some tropical the expanding built environment. cities are expected to experience more precipitation, while others are expected to experience less. In those cities where precipita- Projections for sea level rise are presented in a case study tion is strongly correlated with inter-annual variability, changes format for New York City. Described are the methods used to in the modes with climate change will be critical, and introduce make these projections and the projections themselves. Most of an added element of uncertainty. These cities include Harare, the text in this section is from Horton and Rosenzweig (2010). Melbourne , Delhi , and São Paulo for ENSO, and Athens and London for the NAO . Sea level rise methods It is important to note that in some cases the GCMs have trouble accurately simulating the baseline/observed precipita- The GCM-based methods used to project sea level rise for tion in particular regions. This is especially true in cities that the New York City region include both global (global thermal have strong seasonal precipitation cycles. With a poor handle on expansion and meltwater from glaciers , ice caps, and ice sheets ) the baseline, the projections for the city are therefore skewed. and local (local land subsidence and local water surface eleva- The projections for Delhi, which has a monsoonal climate, pro- tion) components. vide an example of this, as extreme values for increased precipitation may overestimate how wet the climate may become. Within the scientific community, there has been extensive dis- It is therefore important to focus on the direction of change, cussion of the possibility that the GCM approach to sea level rise not so much the actual values. Because it was not possible to may underestimate the range of possible increases, in large part include all cities across the globe, the projections presented because it does not fully consider the potential for land-based here can be used as a proxy for other cities with similar climate ice sheets to melt due to dynamical (motion-related) processes conditions. (Horton et al., 2008 ). For this reason, the NPCC developed an alternative “rapid ice-melt” approach for regional sea level Precipitation projections for Athens, Greece rise projection based on observed trends in melting of the West Antarctic (Velicogna and Wahr, 2006 ) and Greenland ice sheets Athens is one city where annual precipitation is projected to (Rignot and Kanagaratnam, 2006 ) and paleoclimate studies decrease over the next century. In the observed climate record of ice-melt rates during the most recent postglacial period for annual precipitation, there is no statistically signifi cant trend. (Fairbanks, 1989 ). Starting around 20,000 years ago, global Slight decreasing trends in other precipitation variables, such as sea level rose 120 meters and reached nearly present-day levels wet days, have been observed. What appears to be occurring in around 8,000–7,000 years ago. The average rate of sea level rise Greece, as well as many other parts of the globe, is an increase during this ~10,000–12,000 year period was 9.9 to 11.9 cm per in extreme daily precipitation (Nastos and Zerefos, 2007 ). While decade. This information is incorporated into the rapid ice-melt total rainfall and the number of wet days show no change or a scenario projections. decline, when there is precipitation, it is more intense. This could potentially be most damaging, as these rain events can cause The GCM-based sea level rise projections indicate that sea short-duration, fl ash fl ooding. level in New York City may rise by 5 to 13 centimeters in the 70 Urban climate: Processes, trends, and projections

Table 3.5: Sea level rise projections including the rapid ice-melt scenario for New York City.

New York City 2020s 2050s 2080s

Sea level risea +5 to 13 cm +18 to 30 cm +30 to 58 cm

Rapid ice-melt ~13 to 25 cm ~48 to 74 cm ~104 to 140 cm sea level riseb

a Based on seven GCMs and three emissions scenarios. b “Rapid ice-melt scenario” is based on acceleration of recent rates of ice-melt in the Greenland and West Antarctic ice sheets and paleoclimate studies.

The rapid ice-melt sea level rise scenario addresses this possibility. It is based on extrapolation of recent accelerating rates of ice-melt from the Greenland and West Antarctic ice sheets and on paleoclimate studies that suggest sea level rise on the order of 9.9 to 11.9 cm per decade may be possible. Sea level rise projections for New York City in the rapid ice-melt scenario are shown Figure 3.18: New York City sea level rise. Combined observed (black line) and projected sea level rise is shown. Projected model changes through time are applied in the bottom row of Table 3.5. The potential for rapid ice-melt is to the observed historical data. The three thick lines (green, red, and blue) show included in the regional projections for New York City because the average for each emissions scenario across the seven GCMs used for sea level of the great socio-economic consequences should it occur. To rise. Shading shows the central range. The bottom and top lines, respectively, show assess the risk of accelerated sea level rise and climate change each year’s minimum and maximum projections across the suite of simulations. A for the New York City region over the coming years, climate 10-year ﬁ lter has been applied to the observed data and model output. The dotted area between 2002 and 2015 represents the period that is not covered due to the experts need to monitor rates of polar ice-melt, as well as other smoothing procedure. key indicators of global and regional climate change . Source: WCRP, PCMDI, and observed data from NOAA Tides and Currents. 3.5.2.4 Extreme events

Some of the largest climate change effects on cities are asso- 2020s, 18 to 30 centimeters in the 2050s, and 30 to 58 centim- ciated with extreme events, such as heat waves, intense pre- eters in the 2080s. Sea level projections for the three emissions cipitation events, and coastal storms . The frequency, intensity, scenarios agree through the 2040s. Figure 3.18 shows that the and duration of many extreme events are expected to increase B1 scenario produces smaller increases in sea level than the A1B with climate change. The following quantitative New York and A2 scenarios beginning in the 2050s, and only around 2080 City example describes the types of extreme event threats faced does the A2 scenario produce larger values than A1B. The diver- by many cities, although each city will face slightly different gence of A2 from A1B occurs approximately 10 years earlier for extreme event and natural disaster risks. temperature than for sea level rise, in part reflecting the large inertia of the ocean and ice sheets relative to the atmosphere. Following the New York City case, we present an example of qualitative projections for tropical cyclones , an extreme event Sea level rise projections for the New York City region that impacts many cities included in this chapter. It should be are higher than global sea level rise projections (generally by noted that extreme climate events and natural disasters are inter- approximately 15 cm for twenty-fi rst century projections) (IPCC, twined, partly because climate hazards represent a major portion 2007 ). One reason is that the New York metropolitan region is of all natural disasters . Furthermore, climate extremes (such as subsiding by approximately 8 to 10 cm per century. The climate intense precipitation or drought-induced forest fi res) may cause models also have a tendency to produce accelerated sea level secondary natural disasters (such as landslides). The following rise along the northeast US coast, associated in large part with a section is a case of extreme event projections for New York City, projected weakening of the Gulf Stream (Yin et al., 2009 ) . from Horton and Rosenzweig, (2010) and Horton et al., (2010).

The model-based sea level rise projections shown in Table 3.5 Extreme events projections for New York City are characterized by greater uncertainty than the temperature projections, due largely to the possibility that dynamic processes Extremes of temperature and precipitation (with the excep- in polar ice sheets not captured by the GCMs may accelerate tion of drought) tend to have their largest impacts at daily rather melting beyond currently projected levels. This uncertainty is than monthly timescales. Because monthly output from climate weighted towards the upper bound: that is, the probability of sea models is considered more reliable than daily output, simulated level rise lower than that described in the GCM-based projec- changes in monthly temperature and precipitation were calcu- tions in Table 3.5 is very low, and the probability of sea level rise lated; monthly changes through time from each of the 16 GCMs exceeding the GCM projections is relatively high. and three emissions scenarios described earlier in this section 71 Climate change and cities

[ADAPTATION] Box 3.3 Adaptation to sea level rise in Wellington, New Zealand

Chris Cameron, Paul Kos and Nenad Petrovic The latest New Zealand guidance on coastal hazards 2 associated with climate change recommends that councils consider Wellington City Council the impacts of a 0.8 m increase in sea level by 2090. However, given the considerable uncertainty in projections and the pos- INTRODUCTION sibility of catastrophic events, an approach was taken based on testing infrastructure resilience and response via a range Wellington is a coastal city, with an inner harbor and exposed of scenarios. This will allow the development of strategies to southern coast. A pilot study focusing on the impacts of sea manage the expected risk. level rise on a low-lying city suburb has commenced to inform the adaptation approach across the city. Three core scenarios were examined (0.5 m, 1 m, and 2 m) with each having an additional 0.5 m storm surge component Part of the area included in this pilot study (Box Figure 3.3) within the harbour and a 1 m component on the exposed has been identifi ed as a key growth node for urban intensi- southern coast. These scenarios refl ect the most recent sci- fi cation. The area also contains a range of key infrastructure entifi c probabilities in the short term (50–100 years), while including a signifi cant highway, an international airport , utili- allowing for possible higher levels in the longer term. ties, businesses, housing, and community facilities such as schools, pools, libraries, a marina, and a surfi ng beach. Evaluation of the scenarios was carried out in an interdisciplinary cross-council workshop including water, drainage, roading, METHODOLOGY hazards, transport, coastal and recreational, and urban planning experts. For each asset the following information was gathered: Sea level rise was viewed as one of the most critical climate description, ownership, criticality, condition, relocatability, eco- change impacts on the study area, because it lies between nomic value, proposed upgrades. Each asset was then tested only 1 m and 3 m elevation. against each sea level rise scenario to determine potential risks and impacts. Feasible response options were then proposed.

Mapping the scenarios

Sea level rise scenarios were mapped based on ground elevation data from LiDAR (Light Detection And Ranging). The LiDAR data were captured at 1 cm vertical intervals with +/–10 cm accuracy and were veriﬁ ed by ﬁ eld survey. The data were then used to create a digital terrain model. A local vertical datum was developed based on mean high water springs. This allowed for assessment of the highest likely sea level that includes mean tidal elements.

Infrastructure and key existing community facilities were mapped against the sea level rise scenarios so the impacts could be evaluated holistically.

RESULTS

Through qualitative assessment of likely impacts and appropriate responses to the sea level rise scenarios a number of issues were identiﬁ ed: • Degradation of the level of service from the storm-water system • Rising groundwater levels • Need to evaluate response options for at-risk coastal areas across the city • Need for early decision-making for response planning • Interactions and interdependency between assets • Need to prioritise adaptation responses across the city

The workshop highlighted that rising sea levels are likely to Box Figure 3.3: The study area. have some impacts on the storm-water system in the short

2 See www.mfe.govt.nz/publications/climate/preparing-for-coastal-change-guide-for-local-govt/index.html. 72 Urban climate: Processes, trends, and projections

term (next 20–30 years). Some solutions were proffered that stations require power and telecommunications, which must could be developed as part of normal asset management. therefore be maintained through the area at all times.

Low-lying parts of the study area may be susceptible to Similar impacts may occur in other parts of the city, and an increased fl ooding due to a rise in the water-table. This was overall cost–benefi t analysis cannot be completed in isolation regarded as more urgent than “over-topping.” Several likely within a limited area. options for responding were identifi ed and will be consulted on with Councillors and the community. Further detailed work These fi ndings will inform the proposed intensifi cation plans is required to examine the impact, behavior, and response of and other ongoing development, maintenance, and asset groundwater in the study area, together with the likely costs management plans for the area. Findings will also be used for and benefi ts of each response option. discussions within council and the community around priori- tizing, costs, and residual risks. Different responses may be appropriate for the natural dune environment of the southern coast compared to the structur- CONCLUSIONS ally modifi ed northern coast. Maintaining a dune environment on the south coast would help the area retain its high aes- This study gathered and evaluated key information needed thetic and amenity values. Moreover, this could be a more to make an initial assessment of climate change impacts in successful adaptation response, given the adverse effects a localized urban area. It has indicated where further detailed that “hard engineered” structures can have on a beach. work could be undertaken to derive a more accurate assessment of costs and benefi ts. A modifi ed approach based on The importance of taking into consideration linkages between this pilot study will be used across other coastal areas within infrastructure elements was recognised. For example, pumping Wellington City.

[ADAPTATION] Box 3.4 Climate change adaptation in Kokkola, Finland

Philipp Schmidt-Thomé shoreline. Rising sea levels have made the glacial rebound less effective over recent years so that the net land rise has Geological Survey of Finland dropped to about 4–5 mm per year relative to the mean sea level. Normal sea level variation in Kokkola is between –1.0 m Juhani Hannila and +1.5 m relative to mean sea level. The city planning offi ce of Kokkola was thus interested in sea level rise scenarios Technical Service Centre, City of Kokkola for the twenty-fi rst century. The background is that Kokkola experiences a strong pressure for coastal housing develop- Kokkola, a medium-sized town on the west coast of Finland , ment. If the coastal land uplift continued as at present, the was one case study of the Developing Policies & Adaptation coastal areas could be easily developed, even with the lower Strategies to Climate Change in the Baltic Sea Region (ASTRA) glacial rebound effect of 5 mm per year. But what if the sea project . The award-winning ASTRA project was co-fi nanced level rise and storm fl ood events become stronger and fl ood by the European Regional Development Fund (ERDF) under patterns change? the INTERREG IIIB Programme. The ASTRA (www.astra- project.org) and SEAREG (www.gtk.fi /dsf) projects are the Kokkola participated in the ASTRA project to better under- predecessors of the currently ongoing BaltCICA project (www. stand the climate change scenarios and to evaluate how these baltcica.org). The aim of the projects is to support planners can be implemented into local planning. The uncertainties of the and decision-makers in understanding the potential impacts climate change scenarios especially played an important role in of climate change on regional development and to support the assessments. Finally the city chose to use the MPIA2 “high the implementation of adaptation strategies. While SEAREG case” sea level change scenario originally developed under the (2002–2005) mainly focused on awareness raising, ASTRA SEAREG project. According to these scenarios it is possible (2005–2007) went a step further to support the development that the land uplift will be neutralized by sea level rise; in other of adaptation strategies. BaltCICA (2009–2012) builds on the words, the present coastline would not change over the twen- results of the projects by implementing adaptation measures ty-fi rst century. Consequently, the often predicted increasing in cooperation with stakeholders. All three projects are man- wind speed peaks during storm events and the increase in aged by the Geological Survey of Finland (GTK) and count on heavy rainfalls would lead to fl ood-prone area changes. Two partners from the Baltic Sea Region . important locations for future housing development in Kokkola are the area of the 2011 housing fair and Bride Island. Kokkola was founded in 1620 adjacent to the Bothnian Bay (Baltic Sea). The postglacial rebound (land uplift) in this area The area for the housing fair, to be held in 2011, was planned amounts to 9 mm per year, and due to the retreating sea several years ago. Such a housing fair is an important event the original city center now lies about 2 km inland from the in Finland, as the houses built for the fair are later to be used

73 Climate change and cities

for housing, which is an important asset for the investors. The start to threaten housing areas. Due to the scenarios used housing fair in Kokkola is planned in a new housing area on in the ASTRA project, land use plan changes for Bride Island the sea shore. In the course of the ASTRA project the location were put on hold until improved climate change scenarios of the housing fair 2011 was not changed, but the minimum and observed trends have been analyzed. However, sea level elevation of the building ground above mean sea level was change scenarios are taken into account in the minimum ele- raised by 1.0–1.2 m compared to previous plans on the sea vation of buildings above mean sea level, e.g., when old cot- shore and is about 3.5 m (streets 3.0 m) above the mean sea tages are renovated. Lowest building elevation for the newest level. In other words, the decision was taken that sea shore building permits has been raised to 2.5 m above the mean plots may be built up, but it has to be made sure that the sea level – so, in building renovation the adaptation to sea lowest fl oor of the houses is well above potential fl ood levels. level rise is put into practice in small steps or plot by plot. The cost of each plot and house was calculated, including the artifi cial elevation of the building ground, and the investors Recommendations of the ASTRA report were also taken into were willing to accept this extra cost because the demand for account in building and city planning regulations. Several houses located on the sea shore is still rising. regulations are in place to protect from rising wind speeds, cold winds, and storms on the shore. For example, low The second example, Bride Island, is a very popular place houses with optimal roof pitch against wind turbulence on for summer cottages in the close vicinity of the city. The cur- yards; inner courtyards towards the south; plantations and rent trend in many European countries is to improve these fencing against cold wind directions; and also directing main temporary summer homes into cottages that can be used all streets crosswise to the coldest winds to avoid wind tunnel year round, and even to convert them into permanent homes. effects. If the land were to continue to rise out of the sea as it has done so far, such a conversion from a temporary to a perma- The city of Kokkola is not taking part in the ASTRA follow-up nent home would pose no problem. The interest of the land project BaltCICA because, for now, all important decisions owners is not only in converting the houses; the investments on climate change adaptation have been taken or are under certainly would also be justifi ed by rising land and house discussion. Nevertheless, the city stays in close contact with prices. The city of Kokkola, on the other hand, carries respon- the Geological Survey of Finland in order to be informed in a sibilities if land use plans are changed and natural hazards timely manner about the latest research results .

were then applied to the observed daily Central Park record from increase. In contrast, extreme cold events, defined as the number 1971 to 2000 to generate 48 time series of daily data.3 This is of days per year with minimum temperature below 0 °C, are a simplified approach to projections of extreme events, since it expected to become rarer. does not allow for possible changes in the patterns of climate variability through time. However, because changes in vari- Although the percentage increase in annual precipitation is ability for most climate hazards are considered highly uncertain, expected to be relatively small in New York, larger percentage the approach described provides an initial evaluation of how increases are expected in the frequency, intensity, and duration extreme events may change in the future. This level of informa- of extreme precipitation (defined as more than 25 mm, 50 mm, tion with appropriate caveats can assist long-term planners as and 100 mm per day). This projection is consistent both with they begin to prepare adaptation strategies to cope with future theory – a warmer atmosphere is expected to hold more mois- extreme events. ture, and while evaporation is a gradual process, precipitation tends to be concentrated in extreme events – and observed trends Despite their brief duration, extreme events can have large nationally over the twentieth century (Karl and Knight, 1998 ). impacts on cities, so they are a critical component of climate change impact assessment. Table 3.6 indicates how the frequency Due to higher projected temperatures, twenty-fi rst century of heat waves , cold events, intense precipitation, drought, and drought projections reflect the competing influences of more coastal flooding in the New York City region are projected to total precipitation and more evaporation. By the end of the change in the coming decades. The average number of extreme twenty-fi rst century the effect of higher temperatures, espe- events per year for the baseline period is shown, along with the cially during the warm months, on evaporation is expected central 67 percent of the range of the model-based projections. to outweigh the increase in precipitation, leading to more droughts , although the timing and levels of drought projections The total number of hot days, defined as days with a max- are marked by relatively large uncertainty. The rapid increase in imum temperature over 32 or 38 °C, is expected to increase as drought risk through time is reflective of a non-linear response, the twenty-fi rst century progresses. The frequency and dura- because as temperature increases in summer become large, tion of heat waves , defined as three or more consecutive days potential evaporation increases dramatically. Because the New with maximum temperatures above 32 °C, are also expected to York Metropolitan Region has experienced severe multi-year

3 Because they are rare, the drought and coastal storm projections were based on longer time periods. 74 Urban climate: Processes, trends, and projections

Table 3.6: Quantitative changes in extreme events.

Extreme event Baseline (1971–2000) 2020s 2050s 2080s

Number of days/year with maximum temperature exceeding:

~32 °C 14 23 to 29 29 to 45 37 to 64 ~38 °C 0.4a 0.6 to 1 1 to 4 2 to 9

Number of heat waves/yearb 2 3 to 4 4 to 6 5 to 8 Average duration (in days) 4 4 to 5 5 5 to 7

Number of days/year with 72 53 to 61 45 to 54 36 to 49 minimum temperature at or Heat waves and cold events below 0 °C

Number of days per year with 13 13 to 14 13 to 15 14 to 16 rainfall exceeding ~25 mm

c and droughts Drought to occur, on average ~once every 100 ~once every 100 ~once every 50 to ~once every 8 to 100 years years 100 years years Intense precipitation Intense precipitation

d 1-in-10 year ﬂ ood to reoccur, on ~once every 10 years ~once every 8 to 10 ~once every 3 to 6 ~once every 1 to 3 average years years years

Flood heights (m) associated with 1.9 2.0 to 2.1 2.1 to 2.2 2.3 to 2.5 1-in-10 year ﬂ ood

1-in-100 year ﬂ ood to reoccur, on ~once every 100 ~once every 65 to 80 ~once every 35 to 55 ~once every 15 to 35 average years years years years

Flood heights (m) associated with 2.6 2.7 to 2.8 2.8 to 2.9 2.9 to 3.2

Coastal fl oods and storms Coastal fl 1-in-100 year fl ood

The central range (middle 67 percent of values from model-based probabilities) across the GCMs and greenhouse gas emissions scenarios is shown. a Decimal places shown for values less than 1 (and for all ﬂ ood heights). b Deﬁ ned as three or more consecutive days with maximum temperature exceeding ~32 °C. c Based on minima of the Palmer Drought Severity Index (PDSI) over any 12 consecutive months. d Does not include the rapid ice-melt scenario.

droughts during the twentieth century – most notably the 1960s The projections for fl ooding associated with more severe “drought of record” – any increase in drought frequency, inten- storms (e.g., the 1-in-100 year storm) are less well characterized sity, or duration could have serious implications for water than those for less severe storms (e.g., the 1-in-10 year events), for resources in the region. Changes in the distribution of precipita- multiple reasons. The historical record is not suffi ciently long to tion throughout the year, and timing of snow -melt, could poten- allow precise estimates of the fl ood level associated with the once tially make drought more frequent as well. According to the per century storm. Furthermore, the storm risk may vary on multi- IPCC , snow season length is very likely to decrease over North decadal to centennial ocean circulation-driven timescales that are America (IPCC, 2007 ). currently not well understood. Keeping these uncertainties in mind, we estimate that, due to sea level rise alone, the 1-in-100 year flood As sea level rises, coastal flooding associated with storms may occur approximately four times as often by the 2080s. will very likely increase in intensity, frequency, and duration. The changes in coastal floods shown here are solely due to the Tropical cyclones IPCC model-based projections of gradual changes in sea level through time. Any increase in the frequency or intensity of One extreme climate event that impacts many cities around storms themselves would result in even more frequent future the globe is tropical cyclones. Of the cities used in the chapter, flood occurrences. By the end of the twenty-fi rst century, pro- several are at risk of being impacted by these storms, which jections based on sea level rise alone suggest that coastal flood bring heavy rainfall, high winds, and coastal storm surge. King- levels that currently occur on average once per decade may occur ston is one city at risk of tropical cyclones and, in recent years, once every one-to-three years (see Table 3.6 ). several storms have affected the island of Jamaica. These storms 75 Climate change and cities

Figure 3.19: Hurricane tracks near Kingston, Jamaica. Source: Adapted from NOAA. include Ivan (2004), Emily (2005), Dean (2007), and Gustav caused 45 deaths and over US$2 billion in damage (Lawrence (2008). Figure 3.19 shows the tracks of these storms along with and Gross, 1989 ). two others, Charlie (1951) and Gilbert (1988). There is much uncertainty as to how the frequency and Perhaps the most devastating storm to hit the island was Hur- strength of tropical cyclones will change with global climate ricane Gilbert in 1988, which had winds over 50 m/s as it passed change. Patterns of natural climate variability, including El over the island. Heavy rainfall and a storm surge close to 3 m Niño–Southern Oscillation (ENSO) and Atlantic Multidecadal

[VULNERABILITY] Box 3.5 Lessons from a major climate event: Hurricane Katrina and New Orleans

Alexander S. Kolker Katrina’s strength was exacerbated by climate change. The northern Gulf of Mexico is historically hurricane prone, and has Louisiana Universities Marine Consortium, Chauvin, LA experienced large storms throughout the past 3000 years (Liu and Fearn, 2000). The power of Atlantic hurricanes appears to Douglas J. Meffert have increased over the past three decades (Emanuel, 2005), Center for Bioenvironmental Research, Tulane and Xavier though this view is not accepted by all experts (Trenberth and Universities, New Orleans LA Fasullo, 2008). However, climate models generally predict that global warming will increase the severity and/or intensity Armando Carbonell of tropical cyclones (IPCC, 2007), and Hurricane Katrina provides important lessons into how cities choose to adapt to Department of Planning and Urban Form, Lincoln Institute of Land and mitigate future climate change. Policy, Cambridge, MA One lesson is that a city’s physical landscape strongly affects Stephen A. Nelson its response to climate events. New Orleans developed along a series of natural levees at the banks of the Mississippi River Department of Earth and Environmental Sciences, Tulane University, and its former distributaries and the city’s geography can be New Orleans, LA traced to alluvial processes. Rivers deposit their largest and heaviest particles closest to the main channel, and these nat- Few cities in developed countries have felt the impacts of ural levees were the high stable grounds where the city was climatic processes as has New Orleans. On the morning of originally settled (Coleman et al., 1998; Coleman and Prior, August 29, 2005, Hurricane Katrina made landfall in south 1980; Gould, 1970). As the city expanded in the late 19th Louisiana and again in Mississippi. It produced storm surges and 20th centuries, lower-lying areas comprised of muddy, that ruptured levees on drainage and navigation canals organic-rich sediments were developed. These areas were that catastrophically ﬂ ooded the City of New Orleans. Rap- drained, and once dried out, subsided rapidly (Rogers et al., idly rising waters were met by an inadequate governmental 2008; Nelson and LeClair, 2006; Kolb and Saucier, 1992). In response. In the process, well over 1400 people died, and the present city of New Orleans elevation, subsidence rates many more lives were forever changed. and depth of ﬂ ooding following Katrina are roughly propor- tional to age, with younger regions typically being the lower, Katrina was a massive storm, fueled by very warm waters in wetter, and more rapidly subsiding (Dixon et al., 2006; Seed the Gulf of Mexico; however, uncertainty exists as to whether et al., 2008; Russell, 2005).

76 Urban climate: Processes, trends, and projections

New Orleans’s geology also has important implications for These two lessons lead to a third, that climate disasters are the stability of the city’s levees and its vulnerability to sea often the product of an interaction between natural processes level rise. High subsidence rates in the delta plain contrib- and human actions. A smaller storm would have produced uted to one of the highest rates of relative sea level rise on a smaller storm surge and less pressure on the levees while Earth (Tornqvist et al., 2008, Reed, 2002; Day et al., 2007), properly constructed and maintained levees should have been which makes New Orleans a data-rich model for examining able to withstand many of Katrina’s surges in New Orleans. how cities respond to climatically driven sea-level rise. Sub- The catastrophe in New Orleans was exacerbated by inad- sidence lowered elevation of many of the city’s levees while equate local and federal governmental actions that include a and peat and sand bodies below them allowed for subsur- poor assessment of the risks of fl ooding, an inadequate com- face water fl ow that undermined their stability (Rogers et al., munication of these risks, poor levee design, poor levee main- 2008, Seed et al., 2008). Levee instability was also caused tenance, and an inadequate ability to evacuate people and by engineering failures that are discussed below. Human provide for them in times of need (Seed et al., 2008). activities have exacerbated wetland loss in the Mississippi Delta plain, at rates that now stand near 50 km2 yr-1 (Day Looking to the future, New Orleans faces threats and oppor- et al., 2007; Barras et al., 2003; Morton, Benier and Barras, tunities. Global sea levels are predicted to rise, storms may 2006). A simple linear relationship between wetland area increase in severity and frequency, lands surrounding New and storm surge does not exist, as surge magnitude is Orleans will continue to subside and wetland loss is likely affected by factors including the storm’s size, wind speeds, remain problematic. While coastal restoration is needed on a track and the shape of the continental shelf (Chen et al., massive scale, New Orleans fi nds itself hampered by jurisdic- 2008). However, wetland loss has allowed storms surges to tional diffi culties: many coastal restoration decisions are made propagate further inland, increasing the city’s vulnerable to by a range of state, federal and local authorities that some- storms over time. times must balance restoration against other economic or environmental concerns (Carbonell and Meffert, 2009). While it The second lesson from Katrina is simply that the climate may be impossible to restore the entire Mississippi Delta plain, system is capable of delivering vast quantities of energy. The sediment loads are high enough to substantially contribute to maximum wind speed of Hurricane Katrina exceeded 77 m s-1 coastal restoration if managed wisely. Coastal progradation (> 175 mph), the maximum eye wall radius reached 110 km and wetland accretion will likely buffer storm surge and provide and tropical storm force winds extended 370 km from the opportunities for carbon sequestration. Regional groups are storm’s center (McTaggart-Cowan et al., 2007). Storm surges also promoting a “multiple lines of defense” strategy (Lopez, that reached 10.4 m in Biloxi, MS, and ranged between 5.6 – 2006) that views fl ood protection as an integrated system of 6.9 m at Shell Beach east of New Orleans, and 3.95 – 4.75 m natural and man-made components, including barrier islands along the shore of Lake Pontchartrain at the northern edge and beach ridges, wetlands, levees and evacuation plans. of the city (Fritz et al., 2008). Despite the vast power of Hur- New Orleans was settled on high, stable grounds at the mouth ricane Katrina, it is important to recognize that the storm did of the continent’s largest river and this strategic location made not make landfall at New Orleans. This occurred at Buras, LA it desirable for nations looking to establish a claim to the con- and again near Gulfport MS, which are 83 and 68 km from tinent’s interior. Such strategic thinking relating a city to its New Orleans, which was on the west, less intense side of the broader environmental assets and liabilities is key to the future storm (Seed et al., 2008). of this and perhaps other coastal cities.

Oscillation (AMO), have documented relationships with tropical shear , which is dependent on uncertain changes in a range of fac- cyclone activity. For example, when the AMO, a mode of nat- tors (Vecchi and Soden, 2007 ) including the ENSO cycle (Gray, ural variability of sea surface temperatures in the North Atlantic 1984 ; Mann et al. , 2009 ); (2) vertical temperature gradients in Ocean is in the cold phase, hurricane activity increases (Golden- the atmosphere (Emanuel, 2007 ); (3) Saharan dust (Dunion and berg et al. , 2001 ). Velden, 2004 ; Evan et al. , 2006 ); (4) easterly waves and the West African monsoon (Gray, 1979 ; Donnelly and Woodruff, 2007 ); As far as changes in hurricane strength and frequency due to and (5) steering currents, which are infl uenced by a range of fac- anthropogenic climate change go, there is no concrete evidence tors including highly uncertain changes in the NAO (Mann et al. , that global warming is having an infl uence. Although some scien- 2009 ). tifi c studies suggest that warming caused by increased greenhouse gases will increase hurricane intensity (Emanuel, 2005 ), the connection between the two is not conclusive. There are a number 3.6 Conclusions and key research of issues that play into this debate. Sea surface temperatures and questions upper ocean heat content are very likely to increase in the North Atlantic’s main hurricane development regions, and this increase alone will likely favor more intense hurricanes (Emanuel, 2005 ). Climate change is expected to bring warmer temperatures to However, changes in other key factors (not all of which are virtually the entire globe, including all 12 cities analyzed here. mutually exclusive) that infl uence hurricane number and inten- Heat events are projected to increase in frequency, severity and sity are more uncertain. These factors include: (1) vertical wind duration. Total annual precipitation is expected to increase in 77 Climate change and cities some cities, especially in the mid/high latitudes and tropics , Vulnerabilities will also be large among resource-poor popula- and decrease in other cities, especially in the subtropics. Most tions, especially in developing countries, where infrastructure cities are expected to experience an increase in the percentage of may be sub-standard or non-existent, governmental response to their precipitation in the form of intense rainfall events. In many disasters may be limited, and adaptation options may be few for cities, droughts are expected to become more frequent, more reasons including limited capital. severe, and of longer duration. Additionally, rising sea levels are extremely likely in all the coastal cities , and are likely to lead to One implication from this chapter is the need for more more frequent and damaging fl ooding related to coastal storm and improved climate data, especially in cities of developing events in the future. countries. In many cities, the historical record is either too short or the quality too uncertain to support trend analysis and climate Climate change impacts on cities are enhanced by factors change attribution. Without long historical records, the role of including high population density, extensive infrastructure, and climate variability cannot be adequately described, and climate degraded natural environments. Vulnerabilities will be great in change projections will not have as strong a historical footing. many regions that currently experience frequent climate hazards, However, even in cities that have a high quality, lengthy record such as low-lying areas already exposed to frequent fl ooding. of temperature and precipitation, there is a need for additional

[ADAPTATION] Box 3.6 Mexico City’s Virtual Center on Climate Change

Cecilia Conde, Benjamin Martinez and Francisco Estrada The research priorities selected by the City Government are: 1. Assessment of Mexico City air quality, the effect on the Centro de Ciencias de la Atmosfera, UNAM health of people exposed to allergenic bio-particles (pollen), and its relation to climate change Mexico City’s Virtual Center on Climate Change (CVCCCM: 2. Effect of the interaction of temperature and ozone on www.cycccm-atmosfera.unam.mx/cvccccm/) was created Mexico City’s hospital admissions in 2008, with the objectives of: (1) building an entity that 3. The impact of climate change on water availability in the concentrates and organizes the information regarding cli- Metropolitan Area of Mexico City mate change effects on Mexico City, as well as coordinating 4. Vulnerability of potable water sources in Mexico City in research efforts on the subject; (2) supporting the continuous the context of climate change development of public policies that aim to increase adapta- 5. Energy consumption scenarios and emission of green- tive capacity and reduce vulnerability of different social sec- house gases produced by the transport sector in the Met- tors; (3) creating an Adaptation, Vulnerability and Mitigation ropolitan Area of Mexico City Policy Framework for Mexico City. 6. Assessment of the impacts in the Metropolitan Area of Mexico City related to solid waste under climate change The Center aspires to support the development of “useful” conditions science that must answer the questions concerning climate 7. Vulnerability of the ground of conservation of Mexico City variability and change posed by policymakers of the city. to climate change and possible adaptation measures 8. Determination of the vulnerability of the conservation areas City authorities have had to deal with extreme events related of Mexico City to climate change and possible adaptation to the urban heat island, such as heat waves, heavy rains and measures the resulting fl oods, and reduced water availability associated with severe droughts in the catchment basin from which the Of course, several problems have been detected during the city satisfi es part of its water demand. development of this Virtual Center. In particular, “traditional” science has not yet been capable of achieving interdisciplinary Generally, it could be said that cities now experience what research and being stakeholder driven. Currently, researchers could occur under projected climate change scenarios . For are devoted to publishing papers instead of “translating” example, systems outside of Mexico City might be resilient information for decision-making. Most of the policymakers’ to an increase of 1 °C in mean temperatures, but in the city decisions are more focused on resolving the immediate prob- the warming process has already reached more than 3 °C lems than on designing long-term strategies. (Jauregui, 1997). The actors of this Virtual Center are aware of these bar- As a consequence of the uncontrolled growth of the city’s riers, and these problems are being confronted, for example, population and urbanization area, the increase in the occur- through periodic meetings with diverse authorities of Mexico rence of heavy rains since the 1960s has led to an increase in City, which allow direct answers to their questions. This moti- the frequency of severe fl oods. This provides an example of vates a rich discussion from different points of view, resulting Mexico City’s high sensitivity to climate change. in joint strategies that facilitate the incorporation of the found solutions into public policies. It is expected that this critical The Virtual Center must be seen as necessary to face the stage of the research could be fulfi lled during 2010, when climatic problems that could increase in the future. partial results will be presented to Mexico City’s authorities .

78 Urban climate: Processes, trends, and projections station data and climate variables at high temporal resolution to Chen, Q., Wang, L. and Tawes, R. (2008). Hydrodynamic response of improve our understanding of the microclimates that help defi ne northeastern Gulf of Mexico to Hurricanes. Estuaries and Coasts 31, the urban setting and climate risk. If these data can be integrated 1098–1116. Christensen, J.H., et al. (2007) Regional Climate Projection. Climate Change into monitoring systems and real time networks – which can be 2007: The Physical Science Basis. Contribution of Working Group I to an expensive proposition for some cities – weather forecasting the Fourth Assessment Report of the Intergovernmental Panel on Climate can be improved as well. Change, eds. S. Solomon, et al. Cambridge University Press, 849–940. Coleman, J. M., Roberts, H. H. and Stone, G. W. (1998). The Mississippi Furthermore, improved monitoring can help bridge the gaps Delta: An overview. Journal of Coastal Research 14 , 698–716. Coleman, J. M. and Prior, D. B. (1980). Deltaic Sand Bodies. 171. between weather and climate, and climate regions and cities. Collins , M . ( 2005 ). El Nino- or La Nina-like climate change ? Climate This active area of research will enable analysts to use real- Dynamics , 24 , 89 –104. time ocean temperatures to better discern how the frequency of Cunha , L. R . and M . Miller Santos ( 1993 ). The Rio reconstruction project: extreme events (including hurricanes ) may vary by decade in the the fi rst two years. In Towards A Sustainable Urban Environment: The future. As predictions improve, adaptation strategies can be tai- Rio de Janeiro Study , World Bank Discussion Paper 195, Washington, DC, USA: World Bank . lored to refl ect these advances. Monitoring must include impact Day, J. W. et al. (2007). Restoration of the Mississippi Delta: Lessons from variables identifi ed by stakeholders, such as water reservoir Hurricanes Katrina and Rita. Science, 315, 1679–1684. levels, frequency of power failures, and transportation delays. Dixon, T. H. et al. (2006). Subsidence and fl ooding in New Orleans. Nature It is critical that this information not only be collected but also 441, 587–588. stored systematically in a unifi ed database that facilitates the Dilley , M . ( 2000 ). Reducing vulnerability to climate variability in Southern Africa: the growing role of climate information. Climatic Change , 45 , sharing of information and research results across agencies and 63 –73. cities. Donnelly , J. P . and J. D . Woodruff ( 2007 ). Intense hurricane activity over the past 5,000 years controlled by El Nino and the West African monsoon. Given the range of climate hazards and impacts described Nature , 447 , 465 –468. here, there is a critical need for climate change adaptation strate- Dunion , J. P . and C. S . Velden ( 2004 ). The impact of the Saharan air layer on Atlantic tropical cyclone activity . Bulletin of the American Meteoro- gies that align with societal goals such as development, envi- logical Society , 85 , 353 –365. ronmental protection (including greenhouse gas mitigation), and Emanuel , K . ( 2005 ). Increasing destructiveness of tropical cyclones over the social justice and equity. past 30 years . Nature , 436 , 686 –688. Emanuel , K . ( 2007 ). Environmental factors affecting tropical cyclone power dissipation . Journal of Climate , 20 , 5497 –5509. Emanuel , K . ( 2008 ). Hurricanes and global warming: results from downs- REFERENCES caling IPCC AR4 simulations . Bulletin of the American Meteorological Society , 89 , 347 –367. Environment Canada ( 2009 ). Canada’s National Climate and Weather Data Archive. www.climate.weatheroffi ce.gc.ca . Arrhenius , S . (1896 ). On the infl uence of carbonic acid in the air upon Eri , S ., N . Aishali , and L. W . Horowitz ( 2009 ). Present and potential future the temperature of the ground . London, Edinburgh, and Dublin contributions of sulfate, black and organic carbon aerosols from China Philosophical Magazine and Journal of Science ( fi fth series), 41 , to global air quality, premature mortality and radiative . Atmospheric 237 –275. Environment , 43 , 2814– 2822. Australian Bureau of Meteorology ( 2009). Climate information. www.bom. Evan , A. T ., J. P . Dunion , J. A . Foley , A. K . Heidinger , and C. S . Velden gov.au/climate/ . (2006 ). New evidence for a relationship between Atlantic tropical Balling , R. C . and S. W . Brazel ( 1987 ). Diurnal variations in Arizona mon- cyclone activity and African dust outbreaks. Geophysical Research Let- soon precipitation . Monthly Weather Review , 115 , 342– 346. ters , 33 , L19813, doi:19810.11029/12006GL026408. Barras, J. et al. (2003). Historical and Projected Coastal Louisiana Land Fairbanks , R. G . ( 1989 ). 17,000-year glacio-eustatic sea level record: infl u- Changes : 1978–2050. USGS. ence of glacial melting rates on the Younger Dryas event and deep- Bornstein , R . and M . LeRoy ( 1990 ). Urban barrier effects on convective ocean circulation . Nature , 342 , 637 –642. and frontal thunderstorms. Fourth Conference on Mesoscale Processes , Fernandes , E . ( 2000 ). The legalisation of favelas in Brazil: problems and American Meteorological Society , 2. prospects . Third World Planning Review, 22 (2), 167–188. Burian , S. J . and J. M . Shepherd ( 2005 ). Effect of urbanization on the diurnal Fritz, H. M. et al. (2008). Hurricane Katrina storm surge reconnaissance. rainfall pattern in Houston . Hydrological Processes , 19 , 1089 –1103. Journal of Geotechnical and Geoenvironmental Engineering 134, Carbonell, A. and Meffert, D. J. (2009). Climate Change and the Resilience 644–656. of New Orleans: the Adaptation of Deltaic Urban Form . Commissioned G a f fi n , S. R ., C . Rosenzweig , R . Khanbilvardi , et al. ( 2008 ). Variations in Research Report for the World Bank 2009 Urban Research Symposium, New York City’s urban heat island strength over time and space. Theo- Marseilles, France. retical and Applied Climatology , 94 , 1 –11. Census of Canada ( 2006 ). Statistics Canada, accessed 2010, www.statcan. Gao , X ., L . Yong , and W . Lin ( 2003 ). Simulation of effects of land use gc.ca/ . change by a regional climate model . Advances in Atmospheric Sciences , Changnon , S. A . ( 1968 ). The La Port weather anomaly – fact or fi ction ? Bul- 20 , 583 –592. letin of the American Meteorological Society , 49 , 4 –11. Goldenberg , S. B ., C . Landsea , A. M . Mestas - Nunez , and W. M . Gray Changnon , S. A . and N. E. Westcott ( 2002 ). Heavy rainstorms in Chicago: (2001 ). The recent increase in Atlantic hurricane activity: causes and Increasing frequency altered impacts, and future implications . Journal implications . Science , 293 , 474 –479. of American Water Resources , 48 , 1467 –1475. Gould, H. R. (1970), in Deltaic Sedimentation Vol. Special Publication 15, ed. Charlson , R. J ., S. E . Schwartz , J. M . Hales , et al. ( 1992 ). Climate forcing J. P. Morgan. Society of Economic Paleontologists and Mineralogists. by anthropogenic aerosols . Science , 255 , 423 –430. Gray , W. M. (1979 ). Hurricanes: their formation, structure, and likely role in Chen , T. C ., S. Y . Wang , and M. C . Yen ( 2007 ). Enhancement of afternoon the tropical circulation . In D. B . Shaw (Ed.), Meteorology Over the Trop- thunderstorm activity by urbanization in a valley: Tapei. Journal of ical Oceans, London, UK: Royal Meteorological Society, pp. 155–218. Applied Meteorology and Climatology , 46 , 1324 –1340. 79 Climate change and cities

Gray , W. M . ( 1984 ). Atlantic seasonal hurricane frequency. Part I: El Nino IPCC ( 2007 ). Climate Change 2007: The Physical Science Basis. Contribu- and 30 mb quasi-biennial oscillation infl uences . Monthly Weather tion of Working Group I to the Fourth Assessment Report of the Inter- Review , 112 , 1649 –1668. governmental Panel on Climate Change , Cambridge, UK : Cambridge Grimm , A. M . ( 2000 ). Climate variability in southern South America University Press . associated with El Nino and La Nina events. Journal of Climate, 13 , Jáuregui , E . ( 1997 ). Heat island development in Mexico City . Atmospheric 35 –58. Environment , 31 (22), 3821– 3831. Grimm , A. M . ( 2003 ). The El Nino impact on the summer monsoon in Jauregui , E . and E . Romales ( 1996 ). Urban effects on convective precipita- Brazil: regional processes versus remote infl uences . Journal of Climate , tion in Mexico City . Atmospheric Environment , 30 , 3383 –3389. 16 , 263 –280. Jiong, S. (2004). Shanghai’s land use pattern, temperature, relative humidity Grimm , A . ( 2004 ). How do La Niña events disturb the summer monsoon and precipitation. Atlas of Shanghai Urban Physical Geography . system in Brazil ? Climate Dynamics , 22 , 123–138 Karl , T. R . and R. W . Knight ( 1998 ). Secular trends of precipitation amount, Grimm , A. M . and R. G . Tedeschi ( 2009 ). ENSO and extreme rainfall events frequency and intensity in the United States . Bulletin of the American in South America . Journal of Climate , 22 , 1589 –1609. Meteorological Society , 79 , 231 –241. Gupta , P ., S. A . Christopher , J . Wang , et al. (2006 ). Satellite remote sensing Kaufmann , R. K ., K. C . Seto , A . Schneider , et al. ( 2007 ). Climate response of particulate matter and air quality assessment over global cities . to rapid urban growth: evidence of a human-induced precipitation def- Atmospheric Environment , 40 , 5880 –5892. icit . Journal of Climate , 20 , 2299 –2306. Gusmão , P. P ., P . Serrano do Carmo , and S. B . Vianna ( 2008 ). Rio Proximos Kolb, C. R. and Saucier, R. T. (1982). Engineering geology of New Orleans. 100 Anos. Rio De Janiero, Brazil: Instituto Municipal de Urbanismo Review of Engineering Geology 5, 75–93. Pereira Passos . Kothawale , D. R . and K . Rupa Kumar ( 2005 ). On the recent changes in sur- Guttman , N. B . ( 1989 ). Statistical descriptors of climate . Bulletin of the face temperature trends over India . Geophysical Research Letters , 32 . American Meteorological Society , 70 , 602 –607. Kratzer , P . ( 1937 ). Das stadtklima , Braunschweig : F.Viewg uE Sohne . Hachadoorian , L ., S. R . Gaffi n , and R . Engelman ( 2011 ). Mapping the popu- Kratzer , P . ( 1956 ). Das stadtklima (2nd edition), Braunschweig : F Vieweg lation future: projecting a gridded population of the World using ratio uE Sohn (tranlated by the U.S. Air Force, Cambridge Research Labo- methods of trend extrapolation . In R. P. Cincotta and L. J. Gorenfl o ratories). (Eds.), Human Population: The Geography of Homo Sapiens and its Landsberg , H . ( 1956 ). The climate of towns . In W. L . Thomas (Ed.), Man’s Infl uence on Biological Diversity, Heidelberg,Germany: Springer- Role in Changing the Face of the Earth , Chicago, IL, USA : The Uni- Verlag. versity of Chicago Press . Harshvardhan , S. , E . Schwartz , C. M . Benkovitz , and G . Guo ( 2002 ). Aer- Landsberg , H. E . ( 1970 ). Man-made climate changes . Science , 170 , 1265 – osol infl uence on cloud microphysics examined by satellite measure- 1274 ments and chemical transport modeling . Journal of the Atmospheric Lawrence , M. B . and J. M . Gross ( 1989 ). Annual summaries: Atlantic hur- Sciences , 59 , 714 –725. ricane season of 1988 . Monthly Weather Review , 117 , 2248 –2459. Hoerling , M. P ., A . Kumar , and M . Zhong ( 1997 ). El Nino, La Nina and Lian , L . and J . Shu ( 2007 ). Numerical simulation of summer climate over the nonlinearity of their teleconnections. Journal of Climate , 10 , 1769 – center and east China using a regional climate model. Journal of Trop- 1786. ical Meteorology , 23 , 162 –169. Horton , R. E . ( 1921 ). Thunderstorm breeding spots . Monthly Weather Liu, K. B. and Fearn, M. L. (2000). Reconstruction of prehistoric landfall Review , 49 , 193. frequencies of catastrophic hurricanes in northwestern Florida from Horton , R . ( 2007 ). An Observational and Modeling Study of the Regional lake sediment records. Quaternary Research 54, 238–245. Impacts of Climate Variability , Columbia University. Lopez, J. A. (2007). The Multiple Lines of Defense Strategy to Sustain Coastal Horton , R. , C . Herweijer , C . Rosenzweig , et al. ( 2008 ). Sea level rise Louisiana . Lake Pontchartrain Basin Foundation, New Orleans. projects for the current generations CGCMs based on the semi-empir- Mann , M. E ., J. D . Woodruff , J. P . Donnelly , and Z . Zhang ( 2009 ). Atlantic ical method. Geophysical Research Letters , 35 . hurricanes and climate over the past 1,500 years . Nature , 460 , 880 –883. Horton, R., V. Gornitz, M. Bowman, and R. Blake (2010). Climate obser- Marengo , J. A . and C. C . Camargo ( 2008 ). Surface air temperature trends in vations and projections. Ann. New York Acad. Sci., 1196, 41–62, Southern Brazil for 1960–2002. International Journal of Climatology , doi:10.1111/j.1749–6632.2009.05314.x. 28 , 893 –904. Horton, R., and C. Rosenzweig (2010). Climate Risk Information. McTaggart-Cowan, R., Bosart, L. F., Gyakum, J. R. and Altallah, E. H. Ann. New York Acad. Sci., 1196, 147–228, doi:10.1111/j.1749– (2007). Hurricane Katrina (2005) Part 1: Complex life cycle of an 6632.2010.05323.x. intense tropical cyclone. Monthly Weather Review 135, 3905–3926. Howard , L ( 1820 ). The Climate of London, Deduced from Meteorological Morton, R. A., Benier, J. C. and Barras, J. A. (2006). Evidence of regional Observations, Made at Different Places in the Neighbourhood of the subsidence and associated interior wetland loss induced by hydro- Metropolis. 2 , London , 1818 –1820. carbon production, Gulf Coast region, USA. Environmental Geology Huff , F. A . and S. A . Changnon ( 1972a ). Climatological assessment of urban 50, 261–274. effects on precipitation at St. Louis. Journal of Applied Meteorology , Nakicenovic , N . and Coauthors ( 2000 ). Special Report on Emissions Sce- 11 , 823 –842. narios: A Special Report of Working Group III of the Intergovernmental Huff , F. A . and S. A . Changnon ( 1972b ) Climatological Assessment of Panel on Climate Change , Cambridge, UK : Cambridge University Urban Effects on Precipitation St. Louis: Part II. Final Report , NSF Press . Grant GA-18781, Illinois State Water Survey. Nastos , P. T . and C. S . Zerefos ( 2007 ). On extreme daily precipitation totals Hurrell, J. W. (1995). Decadal trends in the North Atlantic Oscillation and at Athens, Greece . Advances in Geosciences , 10 , 59 –66. relationships to regional temperature and precipitation. Science , 269, National Climatic Data Center, Global Historical Climatology Network: 676–679. Version 2 ( 2008). www.ncdc.noaa.gov/oa/climate/ghcn-monthly/index. Hurrel , J. W ., Y . Kushnir , G . Ottersen , and M . Visbeck ( 2003 ). An overview php . of the North Atlantic Oscillation . In J. W . Hurrell et al. (Eds.), The North Nelson, S. A. and LeClair, S. F. (2006). Katrina’s unique splay deposits Atlantic Oscillation: Climatic Signifi cance and Environmental Impact , in a New Orleans neighborhood. GSA Today 16, 4–10, doi: 10.1130/ Geophysical Monograph 134, Washington, DC, USA: American Geo- GSAT01609A.1. physical Union , pp. 1 –35. Ohashi , Y. , Y . Genchi , H . Kondo , et al. ( 2007 ). Infl uence of air-conditioning Ichinose , T . and Y . Bai ( 2000 ). Anthropogenic heat emission in Shanghai waste heat on air temperature in Tokyo during summer: numerical experi- City . In Proceedings of Annual Meeting of Environmental Systems ments using an urban canopy model coupled with a building energy Research, Vol. 28, pp. 329 –337. model . Journal of Applied Meteorology and Climatology , 46 , 66 –81.

80 Urban climate: Processes, trends, and projections

Oke , T. R . ( 1987 ). Boundary Layer Climates (2nd edition) , London, UK : Shepherd , J. M ., H . Pierce , and A. J . Negri ( 2002 ). Rainfall modifi cation Routledge . by major urban areas: observations from spaceborne rain radar on the Orr, J. C., Fabry, V. J., Aumont, O., et al. (2005). Anthropogenic ocean TRMM satellite . Journal of Applied Meteorology , 41 , 689 –701. acidifi cation over the twenty-fi rst century and its impact on calcifying Shu , J ., J. A . Dearing , A. P . Morse , L . Yu , and C . Li ( 2000 ). Magnetic prop- organisms. Nature , 437, 681–686. erties of daily sampled total suspended particulates in Shanghai. Envi- Parson , E ., V . Burkett , K . Fisher-Vanden , et al. (2007 ). Global change sce- ronment Science and Technology , 34 , 2393 –2400. narios: their development and use. Sub-report 2.1B of Synthesis and Shu , J ., J. A . Dearing , A. P . Morse , L . Yu , and N . Yuan ( 2001 ). Deter- Assessment Product 2.1 by the US Climate Change Science Program mining the sources of atmospheric particles in Shanghai, China, from and the Subcommittee on Global Change Research , Washington, DC, magnetic and geochemical properties . Atmospheric Environment , 35 , USA: Department of Energy, Offi ce of Biological and Environmental 2615 –2625. Research, p. 106. Simpson , M. D . ( 2006 ). Role of Urban Land Use on Mesoscale Circulations Pawan , G ., C. A . Sundar , and W . Jun ( 2006 ). Satellite remote sensing of and Precipitation , North Carolina State University. particulate matter and air quality assessment over global cities . Atmos- Stout , G. E . ( 1962 ). Some observations of cloud initiation in industrial pheric Environment , 40 , 5880 –5892. areas. In Air Over Cities , Technical Report A62–5. Washington, DC: Pielke , R. A . ( 2005 ). Land use and climate change . Science , 310 , US Public Health Service. Tayanc, M. and H. Toros (1997). Urbaniza- 1625 –1626. tion effects on regional climate change in the case of four large cities of Reed, D. J. (2002). Sea-level rise and coastal marsh sustainability: geological Turkey . Climate Change , 35 , 501 –524. and ecological factors in the Mississippi delta plain. Geomorphology Tornqvist, T. E. et al. (2008). Mississippi Delta subsidence primarily caused 48, 233–243. by compaction of Holocene strata. Nature Geoscience 1, 173–176. Ren , G. Y ., Z. Y . Chu , Z. H . Chen , and Y. Y . Ren ( 2007 ). Implications Trenberth, K. (1984). Signal versus noise in the Southern Oscillation. of temporal change in urban heat island intensity observed at Bei- Monthly Weather Review 112, 326–332 jing and Wuhan stations , Geophysical Research Letters , 34 , L05711, Trenberth , K. E . and J. M . Caron ( 2000 ). The Southern Oscillation revisited: doi:10.1029/2006GL027927. sea level pressures, surface temperatures and precipitation. Journal of Rignot , E . and P . Kanagaratnam ( 2006 ). Changes in the velocity structure of Climate , 13 , 4358 –4365. the Greenland ice sheet . Science , 311 , 986 –990. Trenberth, K. E. and Fasullo, J. (2008). Energy budgets of Atlantic hurri- Rogers, J. D. et al. (2008). Geological Conditions Underlying the 2005 17th canes and changes from 1970. Geochemistry, Geophysics, Geosystems Street Canal Levee Failure in New Orleans. Journal of Geotechnical 9, doi:2007GC001847. and Geoenvironmental Engineers 134, 583–601. UK Met Offi ce and Hadley Centre (2009). Observations datasets. http:// Ropelewski , C . ( 1999 ). The Great El Nino of 1997 and 1998: impacts on hadobs.org/ . precipitation and temperature . Consequences , 5 , 17 –25. UNFPA ( 2007 ). The State of World Population 2007 . United Nations Popu- Rosenzweig , C ., W. D . Solecki , L . Parshall , and S . Hodges (Eds.) ( 2006 ). lation Fund, United Nations Publications. Mitigating New York City’s Heat Island with Urban Forestry, Living UN-HABITAT ( 2006 ). State of the World’s Cities , London, UK : Earthscan . Roofs, and Light Surfaces, New York City Regional Heat Island Initia- United Nations Statistical Division Demographic Yearbook ( 2010). http:// tive, Final Report 06–06, New York State Energy Research and Devel- unstats.un.org/unsd/demographic/products/dyb/dyb2007.htm . opment Authority. Vecchi , G. A . and B. J . Soden ( 2007 ). Increased tropical Atlantic wind shear Russel, G. (2005) in The Times Picayune Vol. 169 A1 (New Orleans, in model projections of global warming. Geophysical Research Letters , 2005). 34 , L08702, doi:08710.01029/02006GL028905. Sanderson , E . and M . Boyer ( 2009 ). Mannahatta: A Natural History of New Velicogna , I . and J . Wahr ( 2006 ). Acceleration of Greenland ice mass loss in York City , New York, USA : Abrams Books . spring 2004 . Nature , 443 , 329 –331. Satterthwaite , D . ( 2008 ). Cities’ contribution to global warming: notes on Visbeck , M. H ., J. W . Hurrell , L . Polvani , and H. M . Cullen ( 2001 ). The the allocation of greenhouse gas emissions. Journal of Environment and North Atlantic Oscillation: past, present, and future . Proceedings of the Urbanization , 20 , 539 –549. National Academy of Sciences , 98 , 12 ,876–12,877. Seed, R. B. et al. (2008). New Orleans and Hurricane Katrina 1: Introduc- Wilby , R. L ., G . O’Hare , and N . Barnsley ( 1997 ). The North Atlantic Oscil- tion, overview, and the east fl ank. Journal of Geotechnical and Geoen- lation and British Isles climate variability . Weather , 52 , 266 –276. vironmental Engineering 135, 701–739. Yin , J ., M. E . Schlesinger , and R. J . Stouffer ( 2009 ). Model projections of Selover , N . ( 1997 ). Precipitation patterns around an urban desert environ- rapid sea-level rise on the northeast coast of the United States . Nature ment: topographic or urban infl uences ? Association of American Geog- Geoscience , 2 , 262 –266. raphers Convention , 2 . Zhang , X ., L. A . Vincent , W. D . Hogg , and A . Niitsoo ( 2000 ). Temperature Shepherd , J. M . ( 2005 ). A review of current investigations of urban-induced and precipitation trends in Canada during the 20th century . Atmosphere- rainfall and recommendations for the future. Earth Interactions , 9 , Ocean , 38 , 395 –429. 1 –27. Zheng Yiqun , Qian Yongfu , Miao Manqian , et al. ( 2002 ). The effects of Shepherd , J. M . ( 2006 ). Evidence of urban-induced precipitation vari- vegetation change on regional climate I: Simulation results . Acta Mete- ability in arid cliamte regimes. Journal of Arid Environments , 67 , orologica Sinica , 60 , 1 –16 (in Chinese). 607 –628.